正在加载视频...

视频加载失败

返回首页

Breaking news 🗞️ 🚨 Tesla just quietly solved a problem in Australia. The Model X is gone from our market. But the new Model Y L Premium AWD might actually be the closest thing we have to a replacement. And honestly… it makes a lot of sense. ⚡ Tesla...

TeslaGong's profile picture

Tesla in the Gong 🇦🇺🦘🤖🚕

22,748 subscribers

21,519 次观看 • 5 个月前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Woah... the Tesla Semi is no joke. The Tesla team really made it the best semi truck in the world. Just check out the main stats shared from Jay Leno's video. • It goes 500-miles for the long-range model, tested fully loaded and also has a a 325-mile standard-range option available - they both handle max payload • It does 1.2 MW charging, which recovers ~60% of the battery (~300 miles of range) in just 30 minutes during mandatory driver breaks • Has a drag coefficient of ~0.4, which is lower than a Bugatti Veyron bc of the bullet-shaped cab and radical center-driver seating (fyi this is a ~7% improvement over previous design) • The battery is engineered to last 1 million miles, has the same cells as the Cybertruck • It's 50% cheaper to operate in California and nearly 20% cheaper per mile nationwide, this includes everything like energy, maintenance, no oil changes • While Jay Leno was hauling 60-70,000 lbs GVW, he said “I don’t even feel like I’m pulling anything…” • The Tesla Semi fleet already logged >13.5 million miles total, with one truck approaching 440,000 miles with 95% uptime • It's ~1,000 lbs lighter overall + 2,000 lb EV weight exemption keeps it at full payload parity with diesel trucks • Regen braking is so strong it descends steep grades at highway speed with almost zero use of service brakes • The truck is proven in Alaska winters with heat pump and minimal range loss • It has a turning radius like a Model 3/Y despite being a full-size Class 8 truck • Tesla has a new dedicated factory near Giga Reno ramping high-volume production this year (tens of thousands annually) with strong customer demand already locked in If I were a business, the Tesla Semi is a no-brainer!
0:11

Woah... the Tesla Semi is no joke. The Tesla team really made it the best semi truck in the world. Just check out the main stats shared from Jay Leno's video. • It goes 500-miles for the long-range model, tested fully loaded and also has a a 325-mile standard-range option available - they both handle max payload • It does 1.2 MW charging, which recovers ~60% of the battery (~300 miles of range) in just 30 minutes during mandatory driver breaks • Has a drag coefficient of ~0.4, which is lower than a Bugatti Veyron bc of the bullet-shaped cab and radical center-driver seating (fyi this is a ~7% improvement over previous design) • The battery is engineered to last 1 million miles, has the same cells as the Cybertruck • It's 50% cheaper to operate in California and nearly 20% cheaper per mile nationwide, this includes everything like energy, maintenance, no oil changes • While Jay Leno was hauling 60-70,000 lbs GVW, he said “I don’t even feel like I’m pulling anything…” • The Tesla Semi fleet already logged >13.5 million miles total, with one truck approaching 440,000 miles with 95% uptime • It's ~1,000 lbs lighter overall + 2,000 lb EV weight exemption keeps it at full payload parity with diesel trucks • Regen braking is so strong it descends steep grades at highway speed with almost zero use of service brakes • The truck is proven in Alaska winters with heat pump and minimal range loss • It has a turning radius like a Model 3/Y despite being a full-size Class 8 truck • Tesla has a new dedicated factory near Giga Reno ramping high-volume production this year (tens of thousands annually) with strong customer demand already locked in If I were a business, the Tesla Semi is a no-brainer!

Teslaconomics

129,568 次观看 • 4 个月前

🇨🇳 Many people don't know this but Chinese high speed trains are now the fastest in the word and officially faster than even the 🇯🇵 Japanese Shinkansen Chinese trains go 350-400km/h (217-250mph) depending on the model While the Japanese Shinkansen goes 320km/h (200mph) max So I went to try it from Chengdu to Chongqing (just $80 for biz class) This one is the slightly older 350km/h model The business class cabin has 5 big seats that can lay flat with a stewardess dedicated to the cabin who serves you drinks like coffee, water, orange juice etc. I guess if it's a longer trip you probably get food served too The journey was very smooth and very fast, it maintained close to 350km/h at all times and there wasn't many stops or breaks (good) It didn't have WiFi, which is kind of a miss but it seems Chinese just use 5G for everything which worked fine for me without interrruption too I'd say the interior could be a bit more beautiful and modern. I usually say things in Japan look ancient but I think the Shinkansen interior there looks a bit more modern than here The trip took 1 hour and it passed about 300km in that time, roughly equivalent to the width of my country the Netherlands 😂 Anyway solid experience (and cheap!) 👍
0:33

🇨🇳 Many people don't know this but Chinese high speed trains are now the fastest in the word and officially faster than even the 🇯🇵 Japanese Shinkansen Chinese trains go 350-400km/h (217-250mph) depending on the model While the Japanese Shinkansen goes 320km/h (200mph) max So I went to try it from Chengdu to Chongqing (just $80 for biz class) This one is the slightly older 350km/h model The business class cabin has 5 big seats that can lay flat with a stewardess dedicated to the cabin who serves you drinks like coffee, water, orange juice etc. I guess if it's a longer trip you probably get food served too The journey was very smooth and very fast, it maintained close to 350km/h at all times and there wasn't many stops or breaks (good) It didn't have WiFi, which is kind of a miss but it seems Chinese just use 5G for everything which worked fine for me without interrruption too I'd say the interior could be a bit more beautiful and modern. I usually say things in Japan look ancient but I think the Shinkansen interior there looks a bit more modern than here The trip took 1 hour and it passed about 300km in that time, roughly equivalent to the width of my country the Netherlands 😂 Anyway solid experience (and cheap!) 👍

@levelsio

2,012,888 次观看 • 8 个月前

Many still misunderstand Elon and Tesla's mission Elon has a deep concern for humanity, which is why he takes on the toughest problems He's doing it because if he doesn't, nobody will Tesla started the EV revolution... before, nobody even wanted to build EVs. Literally nobody wanted to do it Everyone treated EVs like boring golf carts It was not easy. Being first in the industry means you have to build everything from the ground up, and there is no one you can learn from Tesla got very close to bankruptcy many times... often weeks away Combustion engines had over 150 years of innovation but for EVs, it was the last 15 years that truly made the difference And look at the result: Tesla didn't just build a successful EV; the Tesla Model Y has been the best-selling car in the world for three consecutive years (2023–2025), beating every single gas vehicle on Earth Not to mention, Tesla had to build the entire charging infrastructure from scratch, creating a massive global Supercharger network just so people could drive their cars reliably Other EVs also heavily rely on this network, and Tesla is still helping the competition by opening up their chargers and open-sourcing their patents so anyone can use them for free Without Elon, the world would still be lagging in gas engines
0:16

Many still misunderstand Elon and Tesla's mission Elon has a deep concern for humanity, which is why he takes on the toughest problems He's doing it because if he doesn't, nobody will Tesla started the EV revolution... before, nobody even wanted to build EVs. Literally nobody wanted to do it Everyone treated EVs like boring golf carts It was not easy. Being first in the industry means you have to build everything from the ground up, and there is no one you can learn from Tesla got very close to bankruptcy many times... often weeks away Combustion engines had over 150 years of innovation but for EVs, it was the last 15 years that truly made the difference And look at the result: Tesla didn't just build a successful EV; the Tesla Model Y has been the best-selling car in the world for three consecutive years (2023–2025), beating every single gas vehicle on Earth Not to mention, Tesla had to build the entire charging infrastructure from scratch, creating a massive global Supercharger network just so people could drive their cars reliably Other EVs also heavily rely on this network, and Tesla is still helping the competition by opening up their chargers and open-sourcing their patents so anyone can use them for free Without Elon, the world would still be lagging in gas engines

X Freeze

18,841 次观看 • 3 个月前

Today, I took the new Lucid Gravity SUV out for a quick drive. Here are my initial thoughts of the car and the drive: Materials & Seats: I have to say, this is a nice vehicle. The interior feels very premium, and the materials feel expensive. The perforated vegan leather seats are plush, provide good support, and hug you nicely. I sat in two Gravitys, one specced with the $4,200 Tahoe Brown vegan leather seats (really liked that color) and one with the all-black interior. The massaging seats are also surprisingly useful. Unlike the gimmicky ones in many cars, these actually feel great and not like a tiny mouse poking your back. I’d use them regularly if I owned one. The powered second-row seats are roomy and comfortable, with plenty of space for legs and feet. One small gripe I have: the grab handles on the interior feel like very cheap, hollow plastic. Design: The Gravity leans more toward minivan styling than SUV, but I actually think it works. Lucid managed to fit a whopping 120 cubic feet of cargo capacity into this thing. For context, that’s about 24% more than the Tesla Model X (91.6 cu ft), even though the Gravity is an inch shorter. It’s also just 10% less than the Cadillac Escalade IQ EV (131 cu ft), despite the Escalade being more than two feet longer. The 6K OLED panoramic display looks great, but the software felt a little laggy at times and is tough to see in direct sunlight. The steering wheel button design isn’t my favorite, but I do like the flat-top, flat-bottom design, Reminds me of Cybertruck’s wheel. Cargo loading is very convenient. With the air suspension lowered, the rear load floor is super low, making it easy to get things in and out. The trunk opening, though, is oddly shaped. I found the hatch uncomfortably close to my head, but raising the air suspension helps a little. Taller folks may still run into that issue. Showroom Employees: I met three Lucid employees at the showroom, one of whom had worked at Tesla for 10 years. All of them were very nice and knowledgeable. Sound System: It’s VERY good—great bass response, not muddy. It has Dolby Atmos, and it rocks. The Cybertruck still has the best factory sound system I’ve ever heard, but the Gravity’s is still great. ADAS: Unfortunately, I didn’t have much time to try it out, so I can’t give an opinion. But the 360-degree camera view is nice. It makes parking much easier and has a curb-rash alert so you don’t scrape your expensive wheels. Test Drive: The Gravity I drove, including options, was priced at $115,000. The ride felt firmer than I expected, even in the softest air suspension setting, but it was still comfortable. The sportier modes were fun and deliver that classic, quick-acceleration EV experience. The steering feel struck a good balance—sporty, but not tiring for everyday use. Visibility is good, and the glass windshield that extends past your head is cool (though I still think the Model X windshield is more immersive). Charging: The Gravity has native NACS and can charge at speeds up to 220 kW at Tesla Superchargers. In general, the vehicle is capable of up to 400 kW charging speeds. Final Thoughts: As Tesla fans, we can sometimes be tough on non-Tesla EVs, but I think the Gravity SUV is evidence that Tesla’s mission is working: accelerating the advent of sustainable energy. Tesla’s mission and past work helped pave the way for a vehicle like the Lucid Gravity to exist and come to market. While Lucid’s path to profitability is still in question, and while the Gravity is expensive (for now), even from my short time with it, this feels like Lucid’s first truly great product. Only the $94,900 Grand Touring trim is available right now, but the company says the less expensive Touring trim will come out late this year for $79,900. We shall see. More photos and videos of mine in the thread below:
0:18

Today, I took the new Lucid Gravity SUV out for a quick drive. Here are my initial thoughts of the car and the drive: Materials & Seats: I have to say, this is a nice vehicle. The interior feels very premium, and the materials feel expensive. The perforated vegan leather seats are plush, provide good support, and hug you nicely. I sat in two Gravitys, one specced with the $4,200 Tahoe Brown vegan leather seats (really liked that color) and one with the all-black interior. The massaging seats are also surprisingly useful. Unlike the gimmicky ones in many cars, these actually feel great and not like a tiny mouse poking your back. I’d use them regularly if I owned one. The powered second-row seats are roomy and comfortable, with plenty of space for legs and feet. One small gripe I have: the grab handles on the interior feel like very cheap, hollow plastic. Design: The Gravity leans more toward minivan styling than SUV, but I actually think it works. Lucid managed to fit a whopping 120 cubic feet of cargo capacity into this thing. For context, that’s about 24% more than the Tesla Model X (91.6 cu ft), even though the Gravity is an inch shorter. It’s also just 10% less than the Cadillac Escalade IQ EV (131 cu ft), despite the Escalade being more than two feet longer. The 6K OLED panoramic display looks great, but the software felt a little laggy at times and is tough to see in direct sunlight. The steering wheel button design isn’t my favorite, but I do like the flat-top, flat-bottom design, Reminds me of Cybertruck’s wheel. Cargo loading is very convenient. With the air suspension lowered, the rear load floor is super low, making it easy to get things in and out. The trunk opening, though, is oddly shaped. I found the hatch uncomfortably close to my head, but raising the air suspension helps a little. Taller folks may still run into that issue. Showroom Employees: I met three Lucid employees at the showroom, one of whom had worked at Tesla for 10 years. All of them were very nice and knowledgeable. Sound System: It’s VERY good—great bass response, not muddy. It has Dolby Atmos, and it rocks. The Cybertruck still has the best factory sound system I’ve ever heard, but the Gravity’s is still great. ADAS: Unfortunately, I didn’t have much time to try it out, so I can’t give an opinion. But the 360-degree camera view is nice. It makes parking much easier and has a curb-rash alert so you don’t scrape your expensive wheels. Test Drive: The Gravity I drove, including options, was priced at $115,000. The ride felt firmer than I expected, even in the softest air suspension setting, but it was still comfortable. The sportier modes were fun and deliver that classic, quick-acceleration EV experience. The steering feel struck a good balance—sporty, but not tiring for everyday use. Visibility is good, and the glass windshield that extends past your head is cool (though I still think the Model X windshield is more immersive). Charging: The Gravity has native NACS and can charge at speeds up to 220 kW at Tesla Superchargers. In general, the vehicle is capable of up to 400 kW charging speeds. Final Thoughts: As Tesla fans, we can sometimes be tough on non-Tesla EVs, but I think the Gravity SUV is evidence that Tesla’s mission is working: accelerating the advent of sustainable energy. Tesla’s mission and past work helped pave the way for a vehicle like the Lucid Gravity to exist and come to market. While Lucid’s path to profitability is still in question, and while the Gravity is expensive (for now), even from my short time with it, this feels like Lucid’s first truly great product. Only the $94,900 Grand Touring trim is available right now, but the company says the less expensive Touring trim will come out late this year for $79,900. We shall see. More photos and videos of mine in the thread below:

Sawyer Merritt

110,465 次观看 • 1 年前

Tesla Model YL impressed the heck out of me! I’ve seen all the reviews and noticed many of the same things, so I won’t repeat what others have said, but I want to highlight stuff I haven’t seen many/any people talk about! TL;DR: passenger space is plentiful and not an issue, best model Tesla sells. Game changer for families. Passenger space/comfort I am 6’2 and can comfortably fit in all 3 rows even when adjusting them all to fit me-sized humans! Leg room is 100% not an issue. The only catch is 3rd row headroom. In a proper seating position, my head hit the glass. However, another guy who was 6’0 sat back there and was totally fine, so torso length matters! Even for me, a lil slouch or scooting my butt forward and it was totally fine. 6 adults can easily and comfortably fit in YL for sure. The thigh supports in the front row do absolutely nothing for anyone taller than probably 5’8. Very underwhelming. The captain’s chairs recline REALLY far, almost laying down, combined with the heated AND air conditioned seats and the first two rows are positively luxurious. Arm rests are also great and very sturdy. They come up high enough to still be comfy even for a tall guy like me. Tons of rear vents, air flow was really good and consistent throughout. Trunk space The sub-trunk cover magnetically attaching to the 3rd row seat-back is a great way to claw back some of the lost space when the 3rd row is up. Makes it very usable. Also, people have criticized the bump when the 3rd row is down, but there’s a pair of little folding arms (4th pic) that pops up and makes it totally flat. Great design! Head rests They are adjustable up and down via a mechanical button that you press and hold and just lift/lower the headrest by hand. (3rd video) Pretty decent amount of travel, and a huge improvement! These are on the 1st row AND the captain’s chairs! Camping I had a quick chance to fold all the seats down and climb in on my belly and close the trunk, and this is a huge improvement over the normal Y. I had TONS of room, my toes weren’t touching the trunk and my head wasn’t touching the front seat, not even close! And that was without moving the front seats at all! I sleep in our 2023 Y all the time and it’s great, but it’s only barely enough room and only with the front seats allllll the way forward. Also, the 4 extra vents in the rear cabin will be huge for air flow, which is somewhat of a problem in other Ys. Zen Grey Interior I’m a big proponent of Tesla’s white interior, and I like the Zen Grey! Definitely not stark white, but it’s still very light and looks phenomenal. Doesn’t look dingy at all. If you’re a white interior enthusiast like me, you’ll love Zen Grey too. General observations The showroom was PACKED with people and everyone was there for YL! I talked to a lot of people and let me tell you, families are super excited about YL! It was fun to watch everyone’s reaction as they climbed in the 3rd row and were shocked how usable it is.
0:08

Tesla Model YL impressed the heck out of me! I’ve seen all the reviews and noticed many of the same things, so I won’t repeat what others have said, but I want to highlight stuff I haven’t seen many/any people talk about! TL;DR: passenger space is plentiful and not an issue, best model Tesla sells. Game changer for families. Passenger space/comfort I am 6’2 and can comfortably fit in all 3 rows even when adjusting them all to fit me-sized humans! Leg room is 100% not an issue. The only catch is 3rd row headroom. In a proper seating position, my head hit the glass. However, another guy who was 6’0 sat back there and was totally fine, so torso length matters! Even for me, a lil slouch or scooting my butt forward and it was totally fine. 6 adults can easily and comfortably fit in YL for sure. The thigh supports in the front row do absolutely nothing for anyone taller than probably 5’8. Very underwhelming. The captain’s chairs recline REALLY far, almost laying down, combined with the heated AND air conditioned seats and the first two rows are positively luxurious. Arm rests are also great and very sturdy. They come up high enough to still be comfy even for a tall guy like me. Tons of rear vents, air flow was really good and consistent throughout. Trunk space The sub-trunk cover magnetically attaching to the 3rd row seat-back is a great way to claw back some of the lost space when the 3rd row is up. Makes it very usable. Also, people have criticized the bump when the 3rd row is down, but there’s a pair of little folding arms (4th pic) that pops up and makes it totally flat. Great design! Head rests They are adjustable up and down via a mechanical button that you press and hold and just lift/lower the headrest by hand. (3rd video) Pretty decent amount of travel, and a huge improvement! These are on the 1st row AND the captain’s chairs! Camping I had a quick chance to fold all the seats down and climb in on my belly and close the trunk, and this is a huge improvement over the normal Y. I had TONS of room, my toes weren’t touching the trunk and my head wasn’t touching the front seat, not even close! And that was without moving the front seats at all! I sleep in our 2023 Y all the time and it’s great, but it’s only barely enough room and only with the front seats allllll the way forward. Also, the 4 extra vents in the rear cabin will be huge for air flow, which is somewhat of a problem in other Ys. Zen Grey Interior I’m a big proponent of Tesla’s white interior, and I like the Zen Grey! Definitely not stark white, but it’s still very light and looks phenomenal. Doesn’t look dingy at all. If you’re a white interior enthusiast like me, you’ll love Zen Grey too. General observations The showroom was PACKED with people and everyone was there for YL! I talked to a lot of people and let me tell you, families are super excited about YL! It was fun to watch everyone’s reaction as they climbed in the 3rd row and were shocked how usable it is.

Austin 🇺🇸

195,867 次观看 • 26 天前

This is my "feel the AGI" moment: I used GPT-5.6 Sol to train my own autocorrect model that outperforms GPT-5.6 Sol (wtf??) I have no ML background. I have no idea what I'm doing. I just kept pushing Sol until it spat out a SOTA model. And I spent $0. The motivation: Years of talking to AI have made me terrible at typing. Rather than fix my skill issue, I decided to throw more AI at it. My idea was: instead of autocorrect that interrupts my flow, I want to type fast with mistakes and have AI clean it up after. I wanted the smallest local model possible, for speed, for battery life, for science! So I decided to train my own. Inspired by Andrej Karpathy’s autoresearch, I ran Codex /goal with this setup: pick an experiment, try it, record the results to a doc, throw it out if it fails, and plan the next experiment without repeating failures. I gave a few examples that had to pass, tight latency targets, and let it run. Sol did some amazing things. First, it scanned benchmarks and shortlisted base models: Qwen 3.5, Gemma 4, Liquid LFM 2.5. It found a dataset on HuggingFace for typed text. Then it built a simulator for fingers striking a Mac keyboard, modeling the physical layout with a Gaussian distribution around each key. It simulated striking the wrong key, wrong order, fat-fingering, etc. With the models + data + simulator, it fine-tuned using MLX right on my MacBook. It had a working prototype within an hour! But accuracy was pretty poor. — Problem 1: Tokenization Sol read papers, ran tests, and identified that the tokenizer was the bottleneck. Tokenization makes typos hard for the model to see, so it memorizes mappings instead of using its language priors. Sol tried ByT5, Google’s tokenizer-free byte-level LLM. This made a big improvement, but the model is old and lacked the knowledge needed to reach Sol performance. Sol dug deeper and realized a tokenizer-free model isn’t needed; instead, it used T5Gemma, an encoder-decoder model. This can understand the input deeply before producing output, and furthermore, Sol could post-train the encoder to improve performance. This gave a much higher ceiling. — Problem 2: Loss function Now the model was correcting some typos perfectly, but ignoring most. Sol realized that standard cross-entropy loss was teaching the model to avoid edits, because the vast majority of characters in the training data were left unmodified. The fix was wild: Sol wrote a custom loss function that byte-aligns the source and target strings, uses a dynamic programming algorithm to compute the minimum edits between the two, then weights correct edits much higher than copies. After a lot of tuning, this dramatically improved accuracy. — Problem 3: Autoregression One failure mode remained: if the model made a mistake, it couldn’t backtrack. It could only predict the next token. Teaching it to “think” like a reasoning model would solve this, but would be far too slow. Sol found a beautiful solution: instead of greedily predicting the next token, beam search over all possibilities. This parallelizes the exploration instead of one linear chain-of-thought. At the end, choose the path with highest cumulative log probability. This worked great, but made the experience worse, since the user wouldn’t see progress until the whole search was done. To fix this, Sol made a clever observation: after each search step, the longest common prefix among surviving branches is guaranteed to appear in the final result, so it can be displayed immediately. As the search progresses, weaker paths are dropped and the prefix grows, so the user sees continuous progress. Sol built all this as a custom MLX pipeline that does the parallel decoding on the MacBook GPU, with just ~40ms TTFT. It’s crazy fast and entirely local. — Final eval (error reduction rate, higher is better): - Apple autocorrect: 49.66% - GPT-5.6 Luna: 82.47% - GPT-5.6 Terra: 87.64% - GPT-5.6 Sol: 90.56% - Our model (1.7B): 91.02% Final cost: - 1 quota reset (thanks Tibo) - $0 (And yes, I verified there's no cheating. In fact, we test words scrubbed from the training data to prove the model isn’t memorizing) There were a ton more details and tangents I could write about: contrastive learning, GRPO, DPO, dynamic masking, and more. Sol is a fascinating and creative model. It blew my mind so many times. Don’t let a lack of experience stop you: Sol makes AI experiments accessible to anyone!
0:12

This is my "feel the AGI" moment: I used GPT-5.6 Sol to train my own autocorrect model that outperforms GPT-5.6 Sol (wtf??) I have no ML background. I have no idea what I'm doing. I just kept pushing Sol until it spat out a SOTA model. And I spent $0. The motivation: Years of talking to AI have made me terrible at typing. Rather than fix my skill issue, I decided to throw more AI at it. My idea was: instead of autocorrect that interrupts my flow, I want to type fast with mistakes and have AI clean it up after. I wanted the smallest local model possible, for speed, for battery life, for science! So I decided to train my own. Inspired by Andrej Karpathy’s autoresearch, I ran Codex /goal with this setup: pick an experiment, try it, record the results to a doc, throw it out if it fails, and plan the next experiment without repeating failures. I gave a few examples that had to pass, tight latency targets, and let it run. Sol did some amazing things. First, it scanned benchmarks and shortlisted base models: Qwen 3.5, Gemma 4, Liquid LFM 2.5. It found a dataset on HuggingFace for typed text. Then it built a simulator for fingers striking a Mac keyboard, modeling the physical layout with a Gaussian distribution around each key. It simulated striking the wrong key, wrong order, fat-fingering, etc. With the models + data + simulator, it fine-tuned using MLX right on my MacBook. It had a working prototype within an hour! But accuracy was pretty poor. — Problem 1: Tokenization Sol read papers, ran tests, and identified that the tokenizer was the bottleneck. Tokenization makes typos hard for the model to see, so it memorizes mappings instead of using its language priors. Sol tried ByT5, Google’s tokenizer-free byte-level LLM. This made a big improvement, but the model is old and lacked the knowledge needed to reach Sol performance. Sol dug deeper and realized a tokenizer-free model isn’t needed; instead, it used T5Gemma, an encoder-decoder model. This can understand the input deeply before producing output, and furthermore, Sol could post-train the encoder to improve performance. This gave a much higher ceiling. — Problem 2: Loss function Now the model was correcting some typos perfectly, but ignoring most. Sol realized that standard cross-entropy loss was teaching the model to avoid edits, because the vast majority of characters in the training data were left unmodified. The fix was wild: Sol wrote a custom loss function that byte-aligns the source and target strings, uses a dynamic programming algorithm to compute the minimum edits between the two, then weights correct edits much higher than copies. After a lot of tuning, this dramatically improved accuracy. — Problem 3: Autoregression One failure mode remained: if the model made a mistake, it couldn’t backtrack. It could only predict the next token. Teaching it to “think” like a reasoning model would solve this, but would be far too slow. Sol found a beautiful solution: instead of greedily predicting the next token, beam search over all possibilities. This parallelizes the exploration instead of one linear chain-of-thought. At the end, choose the path with highest cumulative log probability. This worked great, but made the experience worse, since the user wouldn’t see progress until the whole search was done. To fix this, Sol made a clever observation: after each search step, the longest common prefix among surviving branches is guaranteed to appear in the final result, so it can be displayed immediately. As the search progresses, weaker paths are dropped and the prefix grows, so the user sees continuous progress. Sol built all this as a custom MLX pipeline that does the parallel decoding on the MacBook GPU, with just ~40ms TTFT. It’s crazy fast and entirely local. — Final eval (error reduction rate, higher is better): - Apple autocorrect: 49.66% - GPT-5.6 Luna: 82.47% - GPT-5.6 Terra: 87.64% - GPT-5.6 Sol: 90.56% - Our model (1.7B): 91.02% Final cost: - 1 quota reset (thanks Tibo) - $0 (And yes, I verified there's no cheating. In fact, we test words scrubbed from the training data to prove the model isn’t memorizing) There were a ton more details and tangents I could write about: contrastive learning, GRPO, DPO, dynamic masking, and more. Sol is a fascinating and creative model. It blew my mind so many times. Don’t let a lack of experience stop you: Sol makes AI experiments accessible to anyone!

Anshu

178,712 次观看 • 29 天前

my 8 GB VRAM gaming laptop is absolutely going to hate me for this. but I still did it. ran a 31b dense model (Gemma 4 31b Q4) with only 8 GB VRAM last week I ran Gemma 4 26B A4B a mixture of experts model on my RTX 4060 and hit 25–28 tokens/sec using llama.cpp's new MTP support. smooth. snappy. but MoE has a secret: it only activates 4B parameters per token despite having 26B total. that's why it flies. so the real question started haunting me. what if I throw a full, no tricks, every parameter fires on every token, 31B DENSE model at the same machine? # Hardware: GPU: NVIDIA RTX 4060, 8 GB VRAM RAM: 16 GB CPU: Intel Core i7 H Laptop. Gaming. Modest. The model: gemma-4-31B-it-qat-UD-Q4_K_XL.gguf (model's unsloth huggingface link in the comments) This is Google DeepMind's flagship dense model in the Gemma 4 family that can run on single consumer GPU. It packs a hybrid attention architecture, supports up to 256K context natively, and is QAT (Quantization Aware Training) optimized, meaning it retains far more quality than standard post training quants at the same bit depth. This is NOT the MoE. This is 31 BILLION dense parameters, every single one of them loaded. # the flags I used: -m gemma-4-31B-it-qat-UD-Q4_K_XL.gguf -cnv --spec-type draft-mtp --spec-draft-model mtp-gemma-4-31B-it.gguf --spec-draft-n-max 8 --spec-draft-p-min 0.6 -c 6000 -v Multi Token Prediction (MTP) is still active here. Separate draft GGUF required, same as the 26B setup. # Results: → Decode: ~3 tokens/sec → Prefill: ~2 tokens/sec → Context: 6000 tokens → Hardware crying quietly in the corner: yes so is 3 tps actually usable? For real time back and forth chat? Not ideal. You're not having a fluid conversation at 3 tps. but slow ≠ useless. And this is where it gets genuinely interesting. think about how senior devs actually work in a real team. But when something is architectural, deeply complex, or needs serious reasoning? they walk down the hall and escalate to the senior. That's exactly the local AI agent architecture this unlocks: → Fast orchestrator model (Gemma 4 26B MoE at 25+ tps) handles routing, simple queries, tool calls, memory. The junior dev. → Gemma 4 31B dense is the senior, called only when the fast model genuinely hits a wall. Hard multi step reasoning. Complex code generation. Deep architectural decisions. The agentic loop stays fast. Only the hard hops touch the 31B. That's a legitimate production grade local AI architecture on a budget hardware. (requires 2 8gb gpus) other workflows where 3 tps is completely fine: - overnight batch jobs. summarize documents, extract structured data, review code. Fire it off. Sleep. wake up to results. - One shot deep reasoning - Silent code audit loops, you write and test, the 31B reviews diffs and flags issues in the background between your sprints - Any workflow where output quality > output speed A few weeks ago, nobody was running a 30B+ dense model on a single consumer GPU with 8 GB VRAM. At all. Now we're doing it on an Intel i7-H gaming laptop with a NVIDIA RTX 4060, thanks to llama.cpp + QAT quants + MTP speculative drafting. Google DeepMind said the Gemma 4 31B targets "consumer GPUs and workstations." They were not exaggerating. The hardware bar to run serious frontier class models locally keeps dropping. the tools are here. the models are here. you just have to be willing to abuse your laptop a little. what workflows would you actually run on a local 3 tps 31B dense model? genuinely curious. drop it below.
0:18

my 8 GB VRAM gaming laptop is absolutely going to hate me for this. but I still did it. ran a 31b dense model (Gemma 4 31b Q4) with only 8 GB VRAM last week I ran Gemma 4 26B A4B a mixture of experts model on my RTX 4060 and hit 25–28 tokens/sec using llama.cpp's new MTP support. smooth. snappy. but MoE has a secret: it only activates 4B parameters per token despite having 26B total. that's why it flies. so the real question started haunting me. what if I throw a full, no tricks, every parameter fires on every token, 31B DENSE model at the same machine? # Hardware: GPU: NVIDIA RTX 4060, 8 GB VRAM RAM: 16 GB CPU: Intel Core i7 H Laptop. Gaming. Modest. The model: gemma-4-31B-it-qat-UD-Q4_K_XL.gguf (model's unsloth huggingface link in the comments) This is Google DeepMind's flagship dense model in the Gemma 4 family that can run on single consumer GPU. It packs a hybrid attention architecture, supports up to 256K context natively, and is QAT (Quantization Aware Training) optimized, meaning it retains far more quality than standard post training quants at the same bit depth. This is NOT the MoE. This is 31 BILLION dense parameters, every single one of them loaded. # the flags I used: -m gemma-4-31B-it-qat-UD-Q4_K_XL.gguf -cnv --spec-type draft-mtp --spec-draft-model mtp-gemma-4-31B-it.gguf --spec-draft-n-max 8 --spec-draft-p-min 0.6 -c 6000 -v Multi Token Prediction (MTP) is still active here. Separate draft GGUF required, same as the 26B setup. # Results: → Decode: ~3 tokens/sec → Prefill: ~2 tokens/sec → Context: 6000 tokens → Hardware crying quietly in the corner: yes so is 3 tps actually usable? For real time back and forth chat? Not ideal. You're not having a fluid conversation at 3 tps. but slow ≠ useless. And this is where it gets genuinely interesting. think about how senior devs actually work in a real team. But when something is architectural, deeply complex, or needs serious reasoning? they walk down the hall and escalate to the senior. That's exactly the local AI agent architecture this unlocks: → Fast orchestrator model (Gemma 4 26B MoE at 25+ tps) handles routing, simple queries, tool calls, memory. The junior dev. → Gemma 4 31B dense is the senior, called only when the fast model genuinely hits a wall. Hard multi step reasoning. Complex code generation. Deep architectural decisions. The agentic loop stays fast. Only the hard hops touch the 31B. That's a legitimate production grade local AI architecture on a budget hardware. (requires 2 8gb gpus) other workflows where 3 tps is completely fine: - overnight batch jobs. summarize documents, extract structured data, review code. Fire it off. Sleep. wake up to results. - One shot deep reasoning - Silent code audit loops, you write and test, the 31B reviews diffs and flags issues in the background between your sprints - Any workflow where output quality > output speed A few weeks ago, nobody was running a 30B+ dense model on a single consumer GPU with 8 GB VRAM. At all. Now we're doing it on an Intel i7-H gaming laptop with a NVIDIA RTX 4060, thanks to llama.cpp + QAT quants + MTP speculative drafting. Google DeepMind said the Gemma 4 31B targets "consumer GPUs and workstations." They were not exaggerating. The hardware bar to run serious frontier class models locally keeps dropping. the tools are here. the models are here. you just have to be willing to abuse your laptop a little. what workflows would you actually run on a local 3 tps 31B dense model? genuinely curious. drop it below.

Alok

63,689 次观看 • 1 个月前

This one was made with Seedance 2.0 Fast via Dreamina. This is pure Omni-Reference. The only character sheet I used was for these girls, Sari and Ploy. The dude with sarung here and the location were 100% prompted. I didn’t use a character sheet or reference for either of them. Even in Fast mode, Seedance 2.0 is bloody good and it still nails the hyper-vernacular vibe that I always aim for in my work. Seedance 2.0 is both exciting and scary for me 😆 It’s exciting because it is undoubtedly the best model currently available on the market. Trust me, you’ve seen the videos I’ve made so far right? The performance of the model It’s simply the best, period. It has helped me tremendously in creating a shit ton of stories about the region where I live, Southeast Asia. It has been the most exciting thing ever. The scary part is whenever a platform or company comes to me saying, “Hey, we have this new video model. Blah blah blah. We’ll let you know more soon.” It scares the shit out of me because the big question is whether it will be better than Seedance 2.0??? 😆😆 If not, I don’t even want to bother using it. I’ve come this far and achieved this level of quality with Seedance 2.0. That’s why I skipped Happy Horse, which I already tested. It’s also why I’m not bothering with Wan or anything else for now. Their current models are still far inferior to what we already get with Seedance 2.0. I don’t want to downgrade the visual quality. This is also why I need to be really honest. There are certain platforms that host their own in-house models and i'm still part of their CPP. However, because those models are still far behind the quality of Seedance 2.0, I haven’t used them that much. Seedance 2.0 has simply become the benchmark for me. The type of output I’m looking for is also extremely specific, so I can immediately feel it when a model cannot deliver what I need. Seedance 2.0 is definitely not cheap, but it gives me so much creative satisfaction and allows me to make whatever I want. I even have a team that low-key makes softcore erotic videos in the style of Vivamax 😆 I think I’ve trimmed down so many things in my AI workflow because my main goal is to focus on the content itself. If Seedance 2.0 Mini is released soon, I’m dead curious to test it. I think I want to create more stories that revolve around drama rather than highly technical cinematic shots. Seedance 2.0 Fast has been incredibly helpful, but I’m definitely curious to check out the Mini version. But the truth is that I’m completely tool-agnostic. I don’t care which company makes the model. I only care about the quality. You might remember when I praised Grok Imagine Video so damn hard because it was genuinely amazing back then. Then the quality kept getting worse and worse, so I stopped using it. But if it gets better again, I’ll definitely want to use it again. At the end of the day, quality is the only thing that matters.
0:15

This one was made with Seedance 2.0 Fast via Dreamina. This is pure Omni-Reference. The only character sheet I used was for these girls, Sari and Ploy. The dude with sarung here and the location were 100% prompted. I didn’t use a character sheet or reference for either of them. Even in Fast mode, Seedance 2.0 is bloody good and it still nails the hyper-vernacular vibe that I always aim for in my work. Seedance 2.0 is both exciting and scary for me 😆 It’s exciting because it is undoubtedly the best model currently available on the market. Trust me, you’ve seen the videos I’ve made so far right? The performance of the model It’s simply the best, period. It has helped me tremendously in creating a shit ton of stories about the region where I live, Southeast Asia. It has been the most exciting thing ever. The scary part is whenever a platform or company comes to me saying, “Hey, we have this new video model. Blah blah blah. We’ll let you know more soon.” It scares the shit out of me because the big question is whether it will be better than Seedance 2.0??? 😆😆 If not, I don’t even want to bother using it. I’ve come this far and achieved this level of quality with Seedance 2.0. That’s why I skipped Happy Horse, which I already tested. It’s also why I’m not bothering with Wan or anything else for now. Their current models are still far inferior to what we already get with Seedance 2.0. I don’t want to downgrade the visual quality. This is also why I need to be really honest. There are certain platforms that host their own in-house models and i'm still part of their CPP. However, because those models are still far behind the quality of Seedance 2.0, I haven’t used them that much. Seedance 2.0 has simply become the benchmark for me. The type of output I’m looking for is also extremely specific, so I can immediately feel it when a model cannot deliver what I need. Seedance 2.0 is definitely not cheap, but it gives me so much creative satisfaction and allows me to make whatever I want. I even have a team that low-key makes softcore erotic videos in the style of Vivamax 😆 I think I’ve trimmed down so many things in my AI workflow because my main goal is to focus on the content itself. If Seedance 2.0 Mini is released soon, I’m dead curious to test it. I think I want to create more stories that revolve around drama rather than highly technical cinematic shots. Seedance 2.0 Fast has been incredibly helpful, but I’m definitely curious to check out the Mini version. But the truth is that I’m completely tool-agnostic. I don’t care which company makes the model. I only care about the quality. You might remember when I praised Grok Imagine Video so damn hard because it was genuinely amazing back then. Then the quality kept getting worse and worse, so I stopped using it. But if it gets better again, I’ll definitely want to use it again. At the end of the day, quality is the only thing that matters.

DAN · MXVDXN

15,865 次观看 • 1 个月前

🚨12 HOUR NEWS RECAP 1. India launched airstrikes on targets in Pakistan and parts of Kashmir in retaliation to a deadly terrorist attack in Indian-administered Kashmir, which killed 26 tourists. 2. Pakistani authorities claimed the airstrikes hit multiple locations across the country, damaging 4 mosques and a medical clinic, with one official reporting 13 people killed at a Bahawalpur mosque, including women and children. In total, at least 26 people were killed in the strikes. 3. Pakistan offered a conditional ceasefire - but only if India hits the brakes first. Defense Minister Khawaja Asif announced that Islamabad “does not seek war” and is “prepared to exercise restraint.” 4. Trump reportedly plans to make waves next week by officially adopting "Arabian Gulf" or "Gulf of Arabia" as America's designation for the critical waterway long known as the Persian Gulf. 5. Ukraine hit Moscow with drones for the third straight night, shutting down airports just as President Xi landed for Russia’s WW2 Victory Day celebrations. Russia launched missile strikes on Kyiv, killing 2 and injuring several, including 4 children. 6. In his first major interview since leaving office, Biden said that NATO would have to fight Russia for Ukraine to win: “The idea that they could win without engaging NATO in a war, NATO in having engaged the Soviet Union, former Soviet Union, Russia, is not likely.” 7. Tesla rolled out a new long-range rear-wheel drive Model Y in the U.S. for $44,990 - or just $37,490 after the $7,500 federal tax credit. It’s the most affordable long-range Model Y yet, aimed at making premium EVs accessible to more drivers. 8. Flights in and out of Yemen’s Sanaa International Airport were suspended indefinitely following an Israeli airstrike that inflicted extensive damage. The strike also hit a nearby power station, deepening the disruption. 9. Trump didn’t hold back in his latest remarks, branding New York Attorney General Letitia James "a total crook" in response to allegations of mortgage fraud. A recent criminal referral accuses James of falsifying property documents to secure better loan terms. 10. Four former housekeepers are suing Motown legend Smokey Robinson, alleging years of sexual assault, battery, and a toxic work environment enabled by his wife.
0:15

🚨12 HOUR NEWS RECAP 1. India launched airstrikes on targets in Pakistan and parts of Kashmir in retaliation to a deadly terrorist attack in Indian-administered Kashmir, which killed 26 tourists. 2. Pakistani authorities claimed the airstrikes hit multiple locations across the country, damaging 4 mosques and a medical clinic, with one official reporting 13 people killed at a Bahawalpur mosque, including women and children. In total, at least 26 people were killed in the strikes. 3. Pakistan offered a conditional ceasefire - but only if India hits the brakes first. Defense Minister Khawaja Asif announced that Islamabad “does not seek war” and is “prepared to exercise restraint.” 4. Trump reportedly plans to make waves next week by officially adopting "Arabian Gulf" or "Gulf of Arabia" as America's designation for the critical waterway long known as the Persian Gulf. 5. Ukraine hit Moscow with drones for the third straight night, shutting down airports just as President Xi landed for Russia’s WW2 Victory Day celebrations. Russia launched missile strikes on Kyiv, killing 2 and injuring several, including 4 children. 6. In his first major interview since leaving office, Biden said that NATO would have to fight Russia for Ukraine to win: “The idea that they could win without engaging NATO in a war, NATO in having engaged the Soviet Union, former Soviet Union, Russia, is not likely.” 7. Tesla rolled out a new long-range rear-wheel drive Model Y in the U.S. for $44,990 - or just $37,490 after the $7,500 federal tax credit. It’s the most affordable long-range Model Y yet, aimed at making premium EVs accessible to more drivers. 8. Flights in and out of Yemen’s Sanaa International Airport were suspended indefinitely following an Israeli airstrike that inflicted extensive damage. The strike also hit a nearby power station, deepening the disruption. 9. Trump didn’t hold back in his latest remarks, branding New York Attorney General Letitia James "a total crook" in response to allegations of mortgage fraud. A recent criminal referral accuses James of falsifying property documents to secure better loan terms. 10. Four former housekeepers are suing Motown legend Smokey Robinson, alleging years of sexual assault, battery, and a toxic work environment enabled by his wife.

Mario Nawfal

165,214 次观看 • 1 年前

🌟 "SHOULD I BUY AN IPHONE FOR TRACKING?" 🌟 tl;dr at bottom I've been using a facecam and Nvidia tracking for a long time and upgrading to a used iphone 13 combined with vbridger, the difference is HUGE. Here is my take on it! Why is Facecam > iPhone? ✅️ More affordable, esp for those using android phones ✅️ More convenient. If you launch Vtubestudio, there's a setting where your webcam automatically turns on, which is great. ✅️ Can track pretty well in the dark IF you already have a good webcam for night tracking. ❌️❌️ Stiff tracking at times ❌️ Not good at tracking specific mouth movement Why iPhone > Facecam? ✅️✅️ You can make the most of your model, since movement along the X and Y axes are a lot more accurate and wider. Also tracks eyes and overall face better. ✅️ More EXPRESSIONS. If your rigging allows for it, things like cheek puff, and tongue are able to be tracked. As far as I know, I cannot do this on facecam. ❌️❌️ WAY more expensive or requires that you have an iPhone already. Needs more set-up (need phone stand right in front of you, need to hook your phone up to a charger at all times, phone could possibly overheat as well if it's old, so you might need a cooler). TL;DR For me, if you have an extra 200 to spare for WAY better tracking, I would 1000% recommend buying a used iPhone on Amazon. iPhone X is the BARE minumum, I would recommend 12/13 so that your phone does not overheat. I do not need to use a cooler for my used iPhone 13. Being able to use my rigging to its fullest makes the model feel so so so different (in a good way). Feel free to reply with any questions, I will try to answer them!
0:16

🌟 "SHOULD I BUY AN IPHONE FOR TRACKING?" 🌟 tl;dr at bottom I've been using a facecam and Nvidia tracking for a long time and upgrading to a used iphone 13 combined with vbridger, the difference is HUGE. Here is my take on it! Why is Facecam > iPhone? ✅️ More affordable, esp for those using android phones ✅️ More convenient. If you launch Vtubestudio, there's a setting where your webcam automatically turns on, which is great. ✅️ Can track pretty well in the dark IF you already have a good webcam for night tracking. ❌️❌️ Stiff tracking at times ❌️ Not good at tracking specific mouth movement Why iPhone > Facecam? ✅️✅️ You can make the most of your model, since movement along the X and Y axes are a lot more accurate and wider. Also tracks eyes and overall face better. ✅️ More EXPRESSIONS. If your rigging allows for it, things like cheek puff, and tongue are able to be tracked. As far as I know, I cannot do this on facecam. ❌️❌️ WAY more expensive or requires that you have an iPhone already. Needs more set-up (need phone stand right in front of you, need to hook your phone up to a charger at all times, phone could possibly overheat as well if it's old, so you might need a cooler). TL;DR For me, if you have an extra 200 to spare for WAY better tracking, I would 1000% recommend buying a used iPhone on Amazon. iPhone X is the BARE minumum, I would recommend 12/13 so that your phone does not overheat. I do not need to use a cooler for my used iPhone 13. Being able to use my rigging to its fullest makes the model feel so so so different (in a good way). Feel free to reply with any questions, I will try to answer them!

Minori 🎀🍰💢 || bakaneko vampire :3

309,559 次观看 • 1 年前

🚀ASST TO $700 PER SHARE?!?🚀 YOU THINK I'M JOKING? THINK AGAIN, BUCKO. Current ASST snapshot: BTC holdings: 15,000.5 BTC BTC price: $80,593 Bitcoin NAV: $1.21B Total debt: $10M Preferred outstanding: $495.95M Debt + preferred: $505.95M Amplification ratio: 41.9% Current stock price: $15.85 Now here’s the model, and this isn't MOONBOY NONSENSE, kids. This is with Bitcoin at $750k in 2036, not $1 million in 2034. ASST maintains their current 41.9% amplification ratio for 10 years. Translation for normal people: For every $1.00 of Bitcoin NAV, ASST keeps roughly $0.419 of senior claims through debt/preferred financing. The bears hear that and immediately start sweating through a Men’s Wearhouse suit. But this is the actual machine. As Bitcoin rises, the Bitcoin NAV rises. When the NAV rises, the old preferred stack becomes smaller relative to the treasury. So ASST issues more SATA to keep amplification at 41.9%. That new SATA capital buys more Bitcoin. Then Bitcoin goes up again. Then the NAV goes up again. Then the amplification ratio drops again. Then they issue more SATA again. Then they buy more Bitcoin again. This is how you turn a balance sheet into a legally registered orange crocodile. Now we add the funding mix: 75% of new Bitcoin accumulation comes from SATA. 25% comes from issuing common stock. And the common stock is issued at 1.2x EV mNAV. Meaning they are selling equity at a 20% premium to the enterprise value of the Bitcoin stack. That matters. Because issuing common below NAV is financial self-harm. Issuing common above NAV is accretive treasury sorcery. Now assume Bitcoin compounds at 25% per year for 10 years. BTC price goes from: $80,593 today to roughly: $750,579 in year 10 That is a 9.3x move in Bitcoin. Now what happens to ASST? Starting BTC stack: 15,000.5 BTC Projected year 10 BTC stack: 143,425 BTC That is 9.6x more Bitcoin. Starting Bitcoin NAV: $1.21B Projected year 10 Bitcoin NAV: $107.65B That is 89x larger. Now the bears will say: “BUT THE PREFERREDS!” Yes, Carl. The preferreds are the point. Senior claims rise from $505.95M to $45.11B because the model intentionally keeps amplification at 41.9%. That sounds terrifying until you remember the Bitcoin NAV grew to $107.65B. The stack got bigger. The senior claims got bigger. The common equity claim got bigger too. This is where CEBE comes in. CEBE = Common Equity Bitcoin Exposure. It answers the only question that matters: After debt and preferred holders get their claim, how much Bitcoin exposure does the common shareholder really own? Today: Gross BPS: 20,222 sats CEBE/share: 11,759 sats Year 10: Gross BPS: 95,380 sats CEBE/share: 55,416 sats That means common-equity Bitcoin exposure per share rises about 4.7x. Even after common issuance. Even after maintaining the preferred stack. Even after the bears finish their sacred ritual of screaming “DILUTION” into a spreadsheet they opened sideways. Now the share count. Current implied diluted shares: 74.2M Projected year 10 shares: 150.4M So yes, the share count roughly doubles in this model. But the Bitcoin stack goes 9.6x. This is the entire game. If Bitcoin holdings grow much faster than shares outstanding, the common shareholder’s Bitcoin exposure goes up. The bears think all issuance is bad because they learned finance from a Yahoo message board during a divorce. The actual question is: Does issuance increase Bitcoin per share after senior claims? In this model, yes. Now the stock price. Strict 1.2x EV mNAV model gets ASST to about: $559/share But if we anchor the model to today’s actual ASST price of $15.85, the same growth path gets you to roughly: $696/share Call it $700. There it is. ASST to $700 per share is not “vibes.” It is a model. BTC compounds at 25%. SATA funds 75% of accumulation. Common funds 25% at 1.2x EV mNAV. Amplification stays at 41.9%. BTC stack grows from 15,000 BTC to 143,425 BTC. Bitcoin NAV goes from $1.21B to $107.65B. CEBE/share goes from 11,759 sats to 55,416 sats. The stock goes from $15.85 to roughly $700. This is why small Bitcoin treasury companies are so insane. Strategy is the Death Star. ASST is the weird little orange lab experiment in the basement where someone accidentally discovers corporate finance methamphetamine. Tiny denominator. Preferred financing. Bitcoin accumulation. Premium equity issuance. CEBE expansion. A compounding treasury loop. The bear case is that dilution kills the common. The bull case is that accretive dilution plus preferred financing creates a Bitcoin-per-share machine that eats capital markets and leaves behind a pile of traumatized short sellers asking why their model still says “book value.” ASST to $700? If the machine works, yes. If Bitcoin does 25% CAGR, absolutely possible. If SATA scales and common gets issued above NAV, the goblin gets fed. And once the goblin gets fed, the spreadsheet starts looking like it was written by Saylor, Dylan LeClair, and a sleep-deprived Austrian economist locked inside a treasury dashboard with three Celsius energy drinks. This is not financial advice. This is FINANCIAL ENTERTAINMENT:
0:20

🚀ASST TO $700 PER SHARE?!?🚀 YOU THINK I'M JOKING? THINK AGAIN, BUCKO. Current ASST snapshot: BTC holdings: 15,000.5 BTC BTC price: $80,593 Bitcoin NAV: $1.21B Total debt: $10M Preferred outstanding: $495.95M Debt + preferred: $505.95M Amplification ratio: 41.9% Current stock price: $15.85 Now here’s the model, and this isn't MOONBOY NONSENSE, kids. This is with Bitcoin at $750k in 2036, not $1 million in 2034. ASST maintains their current 41.9% amplification ratio for 10 years. Translation for normal people: For every $1.00 of Bitcoin NAV, ASST keeps roughly $0.419 of senior claims through debt/preferred financing. The bears hear that and immediately start sweating through a Men’s Wearhouse suit. But this is the actual machine. As Bitcoin rises, the Bitcoin NAV rises. When the NAV rises, the old preferred stack becomes smaller relative to the treasury. So ASST issues more SATA to keep amplification at 41.9%. That new SATA capital buys more Bitcoin. Then Bitcoin goes up again. Then the NAV goes up again. Then the amplification ratio drops again. Then they issue more SATA again. Then they buy more Bitcoin again. This is how you turn a balance sheet into a legally registered orange crocodile. Now we add the funding mix: 75% of new Bitcoin accumulation comes from SATA. 25% comes from issuing common stock. And the common stock is issued at 1.2x EV mNAV. Meaning they are selling equity at a 20% premium to the enterprise value of the Bitcoin stack. That matters. Because issuing common below NAV is financial self-harm. Issuing common above NAV is accretive treasury sorcery. Now assume Bitcoin compounds at 25% per year for 10 years. BTC price goes from: $80,593 today to roughly: $750,579 in year 10 That is a 9.3x move in Bitcoin. Now what happens to ASST? Starting BTC stack: 15,000.5 BTC Projected year 10 BTC stack: 143,425 BTC That is 9.6x more Bitcoin. Starting Bitcoin NAV: $1.21B Projected year 10 Bitcoin NAV: $107.65B That is 89x larger. Now the bears will say: “BUT THE PREFERREDS!” Yes, Carl. The preferreds are the point. Senior claims rise from $505.95M to $45.11B because the model intentionally keeps amplification at 41.9%. That sounds terrifying until you remember the Bitcoin NAV grew to $107.65B. The stack got bigger. The senior claims got bigger. The common equity claim got bigger too. This is where CEBE comes in. CEBE = Common Equity Bitcoin Exposure. It answers the only question that matters: After debt and preferred holders get their claim, how much Bitcoin exposure does the common shareholder really own? Today: Gross BPS: 20,222 sats CEBE/share: 11,759 sats Year 10: Gross BPS: 95,380 sats CEBE/share: 55,416 sats That means common-equity Bitcoin exposure per share rises about 4.7x. Even after common issuance. Even after maintaining the preferred stack. Even after the bears finish their sacred ritual of screaming “DILUTION” into a spreadsheet they opened sideways. Now the share count. Current implied diluted shares: 74.2M Projected year 10 shares: 150.4M So yes, the share count roughly doubles in this model. But the Bitcoin stack goes 9.6x. This is the entire game. If Bitcoin holdings grow much faster than shares outstanding, the common shareholder’s Bitcoin exposure goes up. The bears think all issuance is bad because they learned finance from a Yahoo message board during a divorce. The actual question is: Does issuance increase Bitcoin per share after senior claims? In this model, yes. Now the stock price. Strict 1.2x EV mNAV model gets ASST to about: $559/share But if we anchor the model to today’s actual ASST price of $15.85, the same growth path gets you to roughly: $696/share Call it $700. There it is. ASST to $700 per share is not “vibes.” It is a model. BTC compounds at 25%. SATA funds 75% of accumulation. Common funds 25% at 1.2x EV mNAV. Amplification stays at 41.9%. BTC stack grows from 15,000 BTC to 143,425 BTC. Bitcoin NAV goes from $1.21B to $107.65B. CEBE/share goes from 11,759 sats to 55,416 sats. The stock goes from $15.85 to roughly $700. This is why small Bitcoin treasury companies are so insane. Strategy is the Death Star. ASST is the weird little orange lab experiment in the basement where someone accidentally discovers corporate finance methamphetamine. Tiny denominator. Preferred financing. Bitcoin accumulation. Premium equity issuance. CEBE expansion. A compounding treasury loop. The bear case is that dilution kills the common. The bull case is that accretive dilution plus preferred financing creates a Bitcoin-per-share machine that eats capital markets and leaves behind a pile of traumatized short sellers asking why their model still says “book value.” ASST to $700? If the machine works, yes. If Bitcoin does 25% CAGR, absolutely possible. If SATA scales and common gets issued above NAV, the goblin gets fed. And once the goblin gets fed, the spreadsheet starts looking like it was written by Saylor, Dylan LeClair, and a sleep-deprived Austrian economist locked inside a treasury dashboard with three Celsius energy drinks. This is not financial advice. This is FINANCIAL ENTERTAINMENT:

Adam Livingston

66,707 次观看 • 3 个月前

⭐️❄️ 1ST CALL SNOW MAP ❄️⭐️ I’m comfortable enough now to put real numbers on a forecast map — not raw model output, but a forecast built from actual meteorological reasoning. This forecast is based on: • Snow ratios • Lower DGZ placement • Front-end focused snowfall • Dry dew points early • Mixing potential near 800 mb (~6,000 ft) as coastal redevelopment occurs 🩶 ZONE 1: 4–6" Lower Manhattan • Long Island • NJ Coast • Philadelphia The front end of this storm should deliver 4–6" of snow, with locally higher amounts, before a transition to sleet and possibly freezing rain. ⚠️ I do NOT see this turning into “just rain.” Ground temperatures will be in the upper teens to lower 20s — meaning freezing rain is a legitimate concern and should not be ignored. 🟦 ZONE 2: 6–10" Interior North/Central NJ • North Shore LI • Upper Boroughs • Lower Westchester • Lower Fairfield CT Soundings support some sleet here, but also a longer-duration front-end snowfall. This is a challenging transition zone — and there is still a realistic scenario where no meaningful changeover occurs due to strong dynamics. ➡️ Higher-end totals are very much on the table. 🟪 ZONE 3: 10–15" NW NJ • Poconos • Lower Hudson Valley • Central CT This area benefits from very high snow ratios for nearly 90% of the storm before eventually shifting to wetter snow. I am very confident that areas in purple do not switch to sleet or freezing rain. This is the jackpot zone. 📌 SIDE THOUGHTS This is my First Call. Adjustments will be made. Forecasting for more than 7+ million homes is never perfect — especially in a setup with this many microclimates and boundary-layer challenges. But this is the best possible forecast with the data we have right now. More updates coming. Stay tuned ON tv UNTIL 10AM PIX11 News ❄️📺
0:15

⭐️❄️ 1ST CALL SNOW MAP ❄️⭐️ I’m comfortable enough now to put real numbers on a forecast map — not raw model output, but a forecast built from actual meteorological reasoning. This forecast is based on: • Snow ratios • Lower DGZ placement • Front-end focused snowfall • Dry dew points early • Mixing potential near 800 mb (~6,000 ft) as coastal redevelopment occurs 🩶 ZONE 1: 4–6" Lower Manhattan • Long Island • NJ Coast • Philadelphia The front end of this storm should deliver 4–6" of snow, with locally higher amounts, before a transition to sleet and possibly freezing rain. ⚠️ I do NOT see this turning into “just rain.” Ground temperatures will be in the upper teens to lower 20s — meaning freezing rain is a legitimate concern and should not be ignored. 🟦 ZONE 2: 6–10" Interior North/Central NJ • North Shore LI • Upper Boroughs • Lower Westchester • Lower Fairfield CT Soundings support some sleet here, but also a longer-duration front-end snowfall. This is a challenging transition zone — and there is still a realistic scenario where no meaningful changeover occurs due to strong dynamics. ➡️ Higher-end totals are very much on the table. 🟪 ZONE 3: 10–15" NW NJ • Poconos • Lower Hudson Valley • Central CT This area benefits from very high snow ratios for nearly 90% of the storm before eventually shifting to wetter snow. I am very confident that areas in purple do not switch to sleet or freezing rain. This is the jackpot zone. 📌 SIDE THOUGHTS This is my First Call. Adjustments will be made. Forecasting for more than 7+ million homes is never perfect — especially in a setup with this many microclimates and boundary-layer challenges. But this is the best possible forecast with the data we have right now. More updates coming. Stay tuned ON tv UNTIL 10AM PIX11 News ❄️📺

Mike Masco

277,811 次观看 • 6 个月前

vPay offshore accounts and physical cards have been getting field-tested IRL for a while now, and we’ll open them to the public as soon as we’re fully confident in the UX. But before offshore accounts go public, I want to address a few points: Some might point out that - vPay isn’t the first crypto card - vPay doesn’t have the lowest fees - So why choose vPay instead of the Coinbase 🛡️ Card or MetaMask 🦊 Card or KAST or or Tria, or any of the other big names? Now to address: Privacy | The biggest differentiator that sets vPay completely apart is Private Banking. The majority of the crypto card providers on the market use Rain infra. Even if you’ve never heard of them, that's what your favorite "NeoBank" uses. And due to their legal jurisdictions, they will report your finances to authorities since they're CRS and FACTA compliant. We are not. As an OmniBank, we work with different banking partners, and although KYC is required to use our services, our offshore banks are non-CRS and non-FACTA. Tax reporting is the responsibility and choice of the user. Offshore Accounts vs Physical Cards | I've tried to highlight this a few times so far. vPay has 3 offerings on the banking side of things. Virtual cards - live now. Physical cards - coming Q1 2026. The first two are similar to what everyone else on the market offers. The offshore accounts are not. which are coming this week. They allow unlimited spending, ATM withdrawals, and international SWIFT transfers, which very few “Neobanks” provide. Offshore accounts are coming this week. Self-Custody | We're not 100% non-custodial yet, as that is near impossible at the moment but it's something we're working towards. And we try to keep the users' self-custodial wallets in the loop as much as possible for maximum control. Those who have tried the vPay app know that almost every move asks for permission from their wallet, and we always encourage users to keep their funds in their non-custodial wallets until the very last moment, since our top-ups usually only take seconds to a minute to process. Fees | All of the card providers mentioned above either raised millions from VCs or in presales or have a huge org backing them. We have neither. vPay was self-funded and community-owned since day 1, launched under Virtuals Protocol Genesis V1 launch model, an objectively bad launch model and hugely unfavorable toward project teams. So even though vPay has been generating revenue and profitable from early on, we do not have the luxury of offering 0% fees yet, since they're mostly a marketing gimmick paid for by millions in VC money and not a sustainable business model for early-stage companies. What we're working towards instead, is true co-ownership of vPay and revenue-share with users. OmniBank vs NeoBank | I’m not a fan of the term “NeoBank.” It implies just a bank, but make it crypto. That’s not vPay. Our goals have always been clear: A) Anything and everything users need to do with their money and assets, both Web2 and Web3, all in one hub. Powered by a constellation of partner agents. The cards and the bank accounts are just the foundation. B) To eventually build independent financial rails for crypto and decouple from the chokehold of Visa/Mastercard. vLink is the first step toward this vision. This turned out to be a rather long tweet, but context matters. Questions and feedback welcome in replies or DMs. See you all with your vPay vCards very soon.
0:08

vPay offshore accounts and physical cards have been getting field-tested IRL for a while now, and we’ll open them to the public as soon as we’re fully confident in the UX. But before offshore accounts go public, I want to address a few points: Some might point out that - vPay isn’t the first crypto card - vPay doesn’t have the lowest fees - So why choose vPay instead of the Coinbase 🛡️ Card or MetaMask 🦊 Card or KAST or or Tria, or any of the other big names? Now to address: Privacy | The biggest differentiator that sets vPay completely apart is Private Banking. The majority of the crypto card providers on the market use Rain infra. Even if you’ve never heard of them, that's what your favorite "NeoBank" uses. And due to their legal jurisdictions, they will report your finances to authorities since they're CRS and FACTA compliant. We are not. As an OmniBank, we work with different banking partners, and although KYC is required to use our services, our offshore banks are non-CRS and non-FACTA. Tax reporting is the responsibility and choice of the user. Offshore Accounts vs Physical Cards | I've tried to highlight this a few times so far. vPay has 3 offerings on the banking side of things. Virtual cards - live now. Physical cards - coming Q1 2026. The first two are similar to what everyone else on the market offers. The offshore accounts are not. which are coming this week. They allow unlimited spending, ATM withdrawals, and international SWIFT transfers, which very few “Neobanks” provide. Offshore accounts are coming this week. Self-Custody | We're not 100% non-custodial yet, as that is near impossible at the moment but it's something we're working towards. And we try to keep the users' self-custodial wallets in the loop as much as possible for maximum control. Those who have tried the vPay app know that almost every move asks for permission from their wallet, and we always encourage users to keep their funds in their non-custodial wallets until the very last moment, since our top-ups usually only take seconds to a minute to process. Fees | All of the card providers mentioned above either raised millions from VCs or in presales or have a huge org backing them. We have neither. vPay was self-funded and community-owned since day 1, launched under Virtuals Protocol Genesis V1 launch model, an objectively bad launch model and hugely unfavorable toward project teams. So even though vPay has been generating revenue and profitable from early on, we do not have the luxury of offering 0% fees yet, since they're mostly a marketing gimmick paid for by millions in VC money and not a sustainable business model for early-stage companies. What we're working towards instead, is true co-ownership of vPay and revenue-share with users. OmniBank vs NeoBank | I’m not a fan of the term “NeoBank.” It implies just a bank, but make it crypto. That’s not vPay. Our goals have always been clear: A) Anything and everything users need to do with their money and assets, both Web2 and Web3, all in one hub. Powered by a constellation of partner agents. The cards and the bank accounts are just the foundation. B) To eventually build independent financial rails for crypto and decouple from the chokehold of Visa/Mastercard. vLink is the first step toward this vision. This turned out to be a rather long tweet, but context matters. Questions and feedback welcome in replies or DMs. See you all with your vPay vCards very soon.

The Dude

20,558 次观看 • 8 个月前

AUKUS, Mistakes and Opportunities In 2016, Japan offered Australia state-of-the-art, diesel-electric, ultra-quiet submarines with the option of local production at the Henderson shipyard. The Australian government rejected the proposal, claiming its goal was always nuclear-powered submarines. Instead, Australia decided to spend roughly A$4-5 billion extending the life of its ageing Collins-class fleet until the 2040s . enough money to have bought seven-eight Japanese Taigei-class submarines outright. If that’s really what the government wanted, the Americans and British certainly sent them the bill for AUKUS. Australia is footing almost the entire cost: A$368 billion over three decades. - The United States receives US$3 billion from Australia to expand its industrial base, build more Virginia-class submarines, and then sells 3–5 second-hand boats back to Canberra. - The United Kingdom receives around £2.4 billion from Australia for design and infrastructure work, shares some development costs, and ends up using the exact same SSN-AUKUS design for its own future fleet at essentially no extra cost. I’m genuinely intrigued by how they managed to sell the Australians on this deal. I’d love to meet and congratulate the American and British negotiators – true sales geniuses. Nuclear submarines must have been a childhood dream of that Australian government; there’s no other explanation. But the problems don’t end there. Just as the Americans have cancelled over 300 programmes and thrown away more than US$200 billion in the last 20 years, the British have serious and very recent issues with their own naval projects. It feels like a structural disease in the Western defence industry. - The Astute programme is more than a decade late, costs have tripled, only 5 of the planned 7 boats have been delivered, and engineering problems keep cropping up. - The Dreadnought class (replacement for the Vanguard ballistic-missile submarines) has ballooned by billions and is now delayed well beyond 2030 because of failures integrating propulsion systems and Trident missiles. - And the crown jewels – the Queen Elizabeth and Prince of Wales aircraft carriers – are operational but chronically short of compatible F-35s and cost a staggering £10 billion in overruns. - The Type 45 destroyers suffered catastrophic electrical failures that left them inoperable for years, and the Type 26 frigate programme has been repeatedly cut back, reflecting completely misplaced priorities. And a programme that is supposed to deliver eight submarines to Australia sometime around 2050–2060 is extremely unlikely to proceed as planned, not only because of budgets and operational complications, but because underwater drones are evolving fast and China is leading that race. The Americans and British have a long naval history, but they are also visionaries who understand perfectly well that the future lies in decentralisation: swarms of UUVs, lithium or solid battery submarines, or even small nuclear-powered ones using micro-reactors. These platforms cost 10–20 % of today’s conventional SSNs to maintain, are lighter, and leave far more internal volume for weapons – meaning smaller, cheaper, and more heavily armed submarines. And what does Australia get left with? Far more than just a submarine partnership with Japan – an entire security ecosystem. By 2026-2028, Japan plans to have the HVPG hypersonic glide vehicles fully operational with ranges up to 2,000 km. Their upgraded Type 12 missiles will reach 1,000–1,500 km and can be launched from ships, aircraft, and land batteries. This is enough to cover and protect the entire Australian coastline for thousands of kilometers. And finally, a 3,000 km-range hypersonic missile is being integrated into the Taigei-class and its successor. That arsenal is far beyond anything currently fielded by any Western nation and only Russia and China have comparable systems.
0:17

AUKUS, Mistakes and Opportunities In 2016, Japan offered Australia state-of-the-art, diesel-electric, ultra-quiet submarines with the option of local production at the Henderson shipyard. The Australian government rejected the proposal, claiming its goal was always nuclear-powered submarines. Instead, Australia decided to spend roughly A$4-5 billion extending the life of its ageing Collins-class fleet until the 2040s . enough money to have bought seven-eight Japanese Taigei-class submarines outright. If that’s really what the government wanted, the Americans and British certainly sent them the bill for AUKUS. Australia is footing almost the entire cost: A$368 billion over three decades. - The United States receives US$3 billion from Australia to expand its industrial base, build more Virginia-class submarines, and then sells 3–5 second-hand boats back to Canberra. - The United Kingdom receives around £2.4 billion from Australia for design and infrastructure work, shares some development costs, and ends up using the exact same SSN-AUKUS design for its own future fleet at essentially no extra cost. I’m genuinely intrigued by how they managed to sell the Australians on this deal. I’d love to meet and congratulate the American and British negotiators – true sales geniuses. Nuclear submarines must have been a childhood dream of that Australian government; there’s no other explanation. But the problems don’t end there. Just as the Americans have cancelled over 300 programmes and thrown away more than US$200 billion in the last 20 years, the British have serious and very recent issues with their own naval projects. It feels like a structural disease in the Western defence industry. - The Astute programme is more than a decade late, costs have tripled, only 5 of the planned 7 boats have been delivered, and engineering problems keep cropping up. - The Dreadnought class (replacement for the Vanguard ballistic-missile submarines) has ballooned by billions and is now delayed well beyond 2030 because of failures integrating propulsion systems and Trident missiles. - And the crown jewels – the Queen Elizabeth and Prince of Wales aircraft carriers – are operational but chronically short of compatible F-35s and cost a staggering £10 billion in overruns. - The Type 45 destroyers suffered catastrophic electrical failures that left them inoperable for years, and the Type 26 frigate programme has been repeatedly cut back, reflecting completely misplaced priorities. And a programme that is supposed to deliver eight submarines to Australia sometime around 2050–2060 is extremely unlikely to proceed as planned, not only because of budgets and operational complications, but because underwater drones are evolving fast and China is leading that race. The Americans and British have a long naval history, but they are also visionaries who understand perfectly well that the future lies in decentralisation: swarms of UUVs, lithium or solid battery submarines, or even small nuclear-powered ones using micro-reactors. These platforms cost 10–20 % of today’s conventional SSNs to maintain, are lighter, and leave far more internal volume for weapons – meaning smaller, cheaper, and more heavily armed submarines. And what does Australia get left with? Far more than just a submarine partnership with Japan – an entire security ecosystem. By 2026-2028, Japan plans to have the HVPG hypersonic glide vehicles fully operational with ranges up to 2,000 km. Their upgraded Type 12 missiles will reach 1,000–1,500 km and can be launched from ships, aircraft, and land batteries. This is enough to cover and protect the entire Australian coastline for thousands of kilometers. And finally, a 3,000 km-range hypersonic missile is being integrated into the Taigei-class and its successor. That arsenal is far beyond anything currently fielded by any Western nation and only Russia and China have comparable systems.

Patricia Marins

86,068 次观看 • 8 个月前

A 27-year-old in Chengdu has been pretending to go to work for 11 months. His mother irons the shirt every morning. Last Tuesday his Polymarket wallet crossed $69,800. He goes by kingofcoinflips. Huawei cut him in June along with the rest of the cloud team. Two months of severance, a stack of recommendation letters nobody opens, and a non-disclosure he keeps in the same drawer as his diploma. He never said a word at home. Every morning he pulls on the same shirt, takes the metro three stops past the Huawei tower, and sits in a co-working desk above a dumpling shop in Chunxi Lu. Six other guys in the same row are running the same routine. Nobody asks. What he brought to the table was 3,400 logged setups since August and a highlighted photocopy of a 1948 Bell Labs paper. Claude Shannon. The MIT professor who quietly compounded 28% a year for three decades and edged out Buffett in the process. He didn't pick stocks. He measured how much information his bets contained, in bits. This kid does the same thing on daily Bitcoin price markets. Every contract gets one number before he touches it: D_KL(P‖Q) = Σ p(x) · log2[p(x)/q(x)] Under 0.05 bits and the fees swallow you. Past 0.10 it's real signal. Past 0.30 your model is broken. Last Tuesday's hit: a BTC Above $80,000 contract quoted at 26.8¢ at sunrise. Order flow on Binance over the prior two hours gave his calculator 0.31 bits. True probability sat closer to 71%. The bot loaded half Kelly. Eight hours later the contract settled at a dollar. +$2,264 on a single click. He stacks a second number on top of every market. How much the Binance tape actually tells him about where Polymarket is heading: I(X;Y) = H(X) - H(X|Y) Under 0.10 bits and the noise wins. Past 0.18 the channel is open. A third loop measures the sharpness of every estimate before sizing: J(θ) = E[(∂/∂θ log f(x|θ))²] When Fisher climbs while mutual information falls, the market is sharpening on noise. He calls it the trap zone. Cost him $9,000 in the first six weeks before he started logging it. Hasn't lost there since. The whole thing exists because Polymarket's daily crypto markets trail the spot tape by roughly forty minutes on slow Asian sessions. Forty minutes is forever for a script and impossible for a person. Out of 3,400 entries, the calculator killed 89% before they ever reached the order book. The 11% that survived built the $69,800 position stack and the $27,500 cumulative since August. His mother sent him a picture of a suit yesterday. Said it's for the family dinner next month. He told her he'd wear it. Plans to tell her the truth at $100,000. His wallet: 99.9% scroll past and call it luck. 0.01% count bits.
0:19

A 27-year-old in Chengdu has been pretending to go to work for 11 months. His mother irons the shirt every morning. Last Tuesday his Polymarket wallet crossed $69,800. He goes by kingofcoinflips. Huawei cut him in June along with the rest of the cloud team. Two months of severance, a stack of recommendation letters nobody opens, and a non-disclosure he keeps in the same drawer as his diploma. He never said a word at home. Every morning he pulls on the same shirt, takes the metro three stops past the Huawei tower, and sits in a co-working desk above a dumpling shop in Chunxi Lu. Six other guys in the same row are running the same routine. Nobody asks. What he brought to the table was 3,400 logged setups since August and a highlighted photocopy of a 1948 Bell Labs paper. Claude Shannon. The MIT professor who quietly compounded 28% a year for three decades and edged out Buffett in the process. He didn't pick stocks. He measured how much information his bets contained, in bits. This kid does the same thing on daily Bitcoin price markets. Every contract gets one number before he touches it: D_KL(P‖Q) = Σ p(x) · log2[p(x)/q(x)] Under 0.05 bits and the fees swallow you. Past 0.10 it's real signal. Past 0.30 your model is broken. Last Tuesday's hit: a BTC Above $80,000 contract quoted at 26.8¢ at sunrise. Order flow on Binance over the prior two hours gave his calculator 0.31 bits. True probability sat closer to 71%. The bot loaded half Kelly. Eight hours later the contract settled at a dollar. +$2,264 on a single click. He stacks a second number on top of every market. How much the Binance tape actually tells him about where Polymarket is heading: I(X;Y) = H(X) - H(X|Y) Under 0.10 bits and the noise wins. Past 0.18 the channel is open. A third loop measures the sharpness of every estimate before sizing: J(θ) = E[(∂/∂θ log f(x|θ))²] When Fisher climbs while mutual information falls, the market is sharpening on noise. He calls it the trap zone. Cost him $9,000 in the first six weeks before he started logging it. Hasn't lost there since. The whole thing exists because Polymarket's daily crypto markets trail the spot tape by roughly forty minutes on slow Asian sessions. Forty minutes is forever for a script and impossible for a person. Out of 3,400 entries, the calculator killed 89% before they ever reached the order book. The 11% that survived built the $69,800 position stack and the $27,500 cumulative since August. His mother sent him a picture of a suit yesterday. Said it's for the family dinner next month. He told her he'd wear it. Plans to tell her the truth at $100,000. His wallet: 99.9% scroll past and call it luck. 0.01% count bits.

Lunar

22,351 次观看 • 3 个月前

VTubing is for everyone! I don't like to bring this up, but recent events in the VTuber community made some people really vile. I had my M&G at HolMat last week and had people come up and just yell at me for being a girl in the "womans world", that is VTubing, no idea about who I was, or why I was there. I had the same group return to my handler multiple times, I felt really worried about the person carrying me around having to deal with this repeatedly and tried to steer them away from the group. It seems they found my YT and left me a handful of the same comments, luckily all caught by moderation tools. While I will refrain from M&G's for a little until this calms down I want to say this; I have been managing for over 1.5 years now, I deliberately take on male VTubers to show them that they can still do it. I take on people that have babiniku accounts, those that are changing from one gender presentation to another and want help, and those that have no gender presentation in their avatar. A lot of male VTubers struggle to find a manager because the stereotype is that all male VTubers are evil and it perpetuates a stereotype that extends beyond entertainment industry subcultures like streaming. VTubing has always been a medium, while character and marketing matters, it's about what makes you happy, YOU are the person that makes the content. Use whatever you want as an avatar, use a voice changer if that's what you want to do. Every month I have a male client bring up they can't do it the same way, that it's easier if jiggle physics or a cute voice is the answer to fast growth. And I tell them yeah, I can also do ASMR, I can do drama content, collab with a larger person, and get to x amount of viewers. There are shortcuts in every entertainment profession and subculture. Does it last? Do quick fixes for anything ever last? Yes, great physics and an expensive model can get people in, but if your value, your way of interacting with people, your content plan and marketing is ass, you can pack up. Most of us start on a small budget, premade or resold models, and it's the same in the big league entertainment industry too. Can I take out a loan and put myself on a billboard tomorrow, can I pay the most expensive model artist and rigger and get in along corporate VTubers tomorrow? Sure. Will it last, will it be genuine, will people trust me? Hell no. Genuine communities and growth build trust. I think streaming, creative industries, entertainment, are full of people, regardless of gender, that will see success and call it "easy", because of what they see as the end product. They don't see most VTubers working a second job, they don't see the managers, they don't see 100+ hours a month going into content production, years invested in singing and voice lessons, model redebut after redebut. I think a lot of male VTubers get a bad rep, because so many boys are raised without putting emphasis on empathy and creativity, watching my brothers be told they should not pursue art, that theatre class is a waste of time, and that they needed to go study x or y to make money for their families in the future was heartbreaking. Nobody should have to look at others and feel so much hatred for society they turn against a whole group of people. If you want to pursue entertainment, please do. If you want to grow and try and make a name for yourself, you should start there, not with yelling about how you already failed. "Oh but I can't", "Oh but the odds are stacked against me", look at the big streamers, look at Ironmouse who overcame everything with hard work, look at Kiara who rose from the ashes, look at everyone that fights against their odds every day and give it your all. If you already think you lost, then you have nothing to lose. Don't give up on your dreams because others tell you to.
0:15

VTubing is for everyone! I don't like to bring this up, but recent events in the VTuber community made some people really vile. I had my M&G at HolMat last week and had people come up and just yell at me for being a girl in the "womans world", that is VTubing, no idea about who I was, or why I was there. I had the same group return to my handler multiple times, I felt really worried about the person carrying me around having to deal with this repeatedly and tried to steer them away from the group. It seems they found my YT and left me a handful of the same comments, luckily all caught by moderation tools. While I will refrain from M&G's for a little until this calms down I want to say this; I have been managing for over 1.5 years now, I deliberately take on male VTubers to show them that they can still do it. I take on people that have babiniku accounts, those that are changing from one gender presentation to another and want help, and those that have no gender presentation in their avatar. A lot of male VTubers struggle to find a manager because the stereotype is that all male VTubers are evil and it perpetuates a stereotype that extends beyond entertainment industry subcultures like streaming. VTubing has always been a medium, while character and marketing matters, it's about what makes you happy, YOU are the person that makes the content. Use whatever you want as an avatar, use a voice changer if that's what you want to do. Every month I have a male client bring up they can't do it the same way, that it's easier if jiggle physics or a cute voice is the answer to fast growth. And I tell them yeah, I can also do ASMR, I can do drama content, collab with a larger person, and get to x amount of viewers. There are shortcuts in every entertainment profession and subculture. Does it last? Do quick fixes for anything ever last? Yes, great physics and an expensive model can get people in, but if your value, your way of interacting with people, your content plan and marketing is ass, you can pack up. Most of us start on a small budget, premade or resold models, and it's the same in the big league entertainment industry too. Can I take out a loan and put myself on a billboard tomorrow, can I pay the most expensive model artist and rigger and get in along corporate VTubers tomorrow? Sure. Will it last, will it be genuine, will people trust me? Hell no. Genuine communities and growth build trust. I think streaming, creative industries, entertainment, are full of people, regardless of gender, that will see success and call it "easy", because of what they see as the end product. They don't see most VTubers working a second job, they don't see the managers, they don't see 100+ hours a month going into content production, years invested in singing and voice lessons, model redebut after redebut. I think a lot of male VTubers get a bad rep, because so many boys are raised without putting emphasis on empathy and creativity, watching my brothers be told they should not pursue art, that theatre class is a waste of time, and that they needed to go study x or y to make money for their families in the future was heartbreaking. Nobody should have to look at others and feel so much hatred for society they turn against a whole group of people. If you want to pursue entertainment, please do. If you want to grow and try and make a name for yourself, you should start there, not with yelling about how you already failed. "Oh but I can't", "Oh but the odds are stacked against me", look at the big streamers, look at Ironmouse who overcame everything with hard work, look at Kiara who rose from the ashes, look at everyone that fights against their odds every day and give it your all. If you already think you lost, then you have nothing to lose. Don't give up on your dreams because others tell you to.

Kuromiya Lucien

15,552 次观看 • 7 个月前

What if any preparations have you seen Iran make ahead of the war? Can you discuss It’s missile capabilities? Any intelligence capabilities? Any surprises it might happen in store? When the 12-day war ended, I estimated that Iran would need about six months to recover, including its nuclear program. Contrary to popular belief, Iran did not lose its entire long-range or medium-range air defense network; while some launchers were damaged, the primary targets of the Israeli strikes were the radar systems. Once the radars were neutralized, Iran successfully hid the bulk of its remaining batteries, leaving much of its arsenal intact. In contrast, short-range systems like the Tor-M1 and domestic variants were heavily engaged against cruise missiles, often being lost or damaged only after their ammunition was completely exhausted. Since then, Iran has worked to rebuild its destroyed radar network and, above all, to implement a genuine counterintelligence doctrine. The Mossad operations against Iranian radars and air defense systems have shaped new perimeter defense and counterintelligence doctrines not only in Iran but in other countries as well. If we look at the quantity of weapons and the organization of armed groups during Iran’s most recent protests, I would say the problem of foreign intelligence operations inside the country remains severe. This seriously threatens much of Iran’s capabilities, and I foresee a wave of sabotage operations as a new war draws closer. Iran has begun receiving collaboration from China across multiple areas,from satellites to internal counterintelligence, but it may still take some time for this to produce tangible results. During the last years, the Mossad relied heavily on cell phones, using SMS for recruitment and accessing device GPS for target location. Iran has since focused intensely on preventing any repetition of this, and on this specific issue, the Chinese appear to have provided support. Although foreign intelligence services have operated extensively inside Iran, the scale of any armed opposition groups is negligible compared to the Iranian armed forces, which could still draw on allied paramilitaries and militias in neighboring countries, including the Houthis. Iran has become a missile power with a stockpile far larger than Western estimates suggest. As early as 1998, Iran was already producing missiles with ranges exceeding 1,000 km, and it has continued doing so ever since, developing 12 to 15 different models in that range - meaning all are capable of reaching Israel. That is nearly 30 years of continuous missile production, resulting in a stockpile of several thousands. Another area where Iran has emerged as a global power is drones, including underwater ones. Iran’s UUVs have evolved rapidly into mass-produced models with integrated AI, and I believe they hold some major surprises in reserve. A key point today is that the AN/TPY-2 radars, which played a critical role in tracking Iranian missiles, would be among the first targets to be engaged. These high-powered X-band radars are the backbone of regional missile defense, providing essential data to THAAD and Patriot batteries. However, because they are large, stationary, and emit high-energy signals, they are highly vulnerable to a first-strike or saturation attack, which would effectively 'blind' the entire defensive network. Obviously, a defense budget of nearly one trillion dollars cannot be compared to Iran’s, but the real question is whether the cost and effort are worth the potential casualties. Even without Israel, the Americans maintain an immense advantage in aerial operations over Iran; however, as I have stated before, this superiority does not translate to the maritime theater.
0:18

What if any preparations have you seen Iran make ahead of the war? Can you discuss It’s missile capabilities? Any intelligence capabilities? Any surprises it might happen in store? When the 12-day war ended, I estimated that Iran would need about six months to recover, including its nuclear program. Contrary to popular belief, Iran did not lose its entire long-range or medium-range air defense network; while some launchers were damaged, the primary targets of the Israeli strikes were the radar systems. Once the radars were neutralized, Iran successfully hid the bulk of its remaining batteries, leaving much of its arsenal intact. In contrast, short-range systems like the Tor-M1 and domestic variants were heavily engaged against cruise missiles, often being lost or damaged only after their ammunition was completely exhausted. Since then, Iran has worked to rebuild its destroyed radar network and, above all, to implement a genuine counterintelligence doctrine. The Mossad operations against Iranian radars and air defense systems have shaped new perimeter defense and counterintelligence doctrines not only in Iran but in other countries as well. If we look at the quantity of weapons and the organization of armed groups during Iran’s most recent protests, I would say the problem of foreign intelligence operations inside the country remains severe. This seriously threatens much of Iran’s capabilities, and I foresee a wave of sabotage operations as a new war draws closer. Iran has begun receiving collaboration from China across multiple areas,from satellites to internal counterintelligence, but it may still take some time for this to produce tangible results. During the last years, the Mossad relied heavily on cell phones, using SMS for recruitment and accessing device GPS for target location. Iran has since focused intensely on preventing any repetition of this, and on this specific issue, the Chinese appear to have provided support. Although foreign intelligence services have operated extensively inside Iran, the scale of any armed opposition groups is negligible compared to the Iranian armed forces, which could still draw on allied paramilitaries and militias in neighboring countries, including the Houthis. Iran has become a missile power with a stockpile far larger than Western estimates suggest. As early as 1998, Iran was already producing missiles with ranges exceeding 1,000 km, and it has continued doing so ever since, developing 12 to 15 different models in that range - meaning all are capable of reaching Israel. That is nearly 30 years of continuous missile production, resulting in a stockpile of several thousands. Another area where Iran has emerged as a global power is drones, including underwater ones. Iran’s UUVs have evolved rapidly into mass-produced models with integrated AI, and I believe they hold some major surprises in reserve. A key point today is that the AN/TPY-2 radars, which played a critical role in tracking Iranian missiles, would be among the first targets to be engaged. These high-powered X-band radars are the backbone of regional missile defense, providing essential data to THAAD and Patriot batteries. However, because they are large, stationary, and emit high-energy signals, they are highly vulnerable to a first-strike or saturation attack, which would effectively 'blind' the entire defensive network. Obviously, a defense budget of nearly one trillion dollars cannot be compared to Iran’s, but the real question is whether the cost and effort are worth the potential casualties. Even without Israel, the Americans maintain an immense advantage in aerial operations over Iran; however, as I have stated before, this superiority does not translate to the maritime theater.

Patricia Marins

21,142 次观看 • 5 个月前

You Can't Vibe-Code Trust Avishai Abrahami, Co-Founder & CEO of Wix , interviewed by Harry Stebbings (kevin andres) Summary: Wix trades at a $2.8B market cap on $2.1B of revenue while the market ascribes roughly zero value to a business throwing off $400M a year in free cash flow. Wix CEO Avishai Abrahami's argument is that the market can't yet price what AI actually threatens: the moat is trust and business logic, and neither gets vibe-coded away. His response is to own the disruptor (Base44), train his own narrow models, and stay committed through a storm he insists always arrives on a random Wednesday. 1. Trust is the moat. The real value of Salesforce is trust: JP Morgan and huge banks let it hold all their customer data, and the CRM itself is a small part of that. "What other platform will JP Morgan trust for their customers' data? None." That trust took years to build and can't be reconstructed by an agent scraping a database, so the companies whose value lives in trust survive the SaaS apocalypse while the ones reduced to piping get commoditized. 2. The business-logic wall. "You're not going to vibe-code Shopify no matter how good you are. The business logic is too hard." Wix tested this directly: they asked a team of professional developers to build the operating logic for a single hairdresser in Base44, gave up after a week, brought in a stronger team, and still failed two weeks later. Complex operational software is far harder than a demo suggests, which is why the pizza shop and the hairdresser stay Wix customers rather than build their own stack. 3. Own the disruptor. Wix bought Base44, a one-person company, for $80M, and it now does over $150M in ARR, roughly double what they paid. Abrahami frames the future as three buckets: owners who never want to build, owners who vibe-code everything themselves, and a mix in the middle over the next five or six years. Rather than bet on which wins, Wix owns the tool customers would defect to, so a customer who switches platforms still switches to Wix. 4. Trading on someone else's news. "Today we are trading on other companies' news. We're not trading on Wix news. We're trading on what OpenAI or Anthropic or Google are saying." Base44 alone, valued on vibe-coding peer multiples, should be worth around $8B, which means the market assigns less than zero to Wix's core. Abrahami's response is to detach: he doesn't wake up checking whether the stock moved 20%, because the only thing he can influence is the business. 5. The narrow model. Wix fine-tuned and combined its own models and now matches top-tier frontier quality on Base44 tasks at far lower cost. The logic: they sit on a huge stream of training data from watching what users try and where they fail, so a model built for Base44 can skip what frontier models carry, like knowledge of Chinese poetry, and go deep on what someone means when they say "build me a task manager to tell my boyfriend where he's wrong." A narrow target is easier to hit than a frontier model, and Wix already runs a trained model on website generation that's faster, cheaper, and makes fewer errors, retrained weekly on a live feedback loop. 6. Quality before cost. When Harry cites Chamath's claim that open source runs 14-16x cheaper, Abrahami pushes back: that holds for small tasks, but for something as complex as Base44 the savings land at 5-10%, and his own model runs 1-30% cheaper than frontier, not the order of magnitude people assume. More to the point, this is the wrong time to chase cost: "20% more quality, 20% less cost, I'll go for the quality." It's a brand-new market that's just starting, and the job now is to make the product better. 7. The but is very big. "We all give too much credit for AI. It's amazing, it's incredible, it's super powerful, but the but is pretty big." He asked Claude to write a safety protocol and got six mandatory gates, then pushed back on each one and watched the model cave until only one survived, downgrading the rest from "must test" to "might want to look at later." We over-trust these systems, and that reflex, treating a Reddit post as equivalent to research published in Nature, is where the danger lives. 8. Customer support still breaks. Wix has 3,500 people and its single biggest department is customer support, serving 192 countries. They tried hard not to build their own AI support agent, tested many off-the-shelf products, and concluded flatly: "It doesn't work. We tried, we tried again, it didn't work." The gap between hyped AI support startups and what actually ships in production is the tell that the technology is earlier than the marketing, maybe five years from being different. 9. Buybacks as dividends. Wix had $1.5B sitting in the bank it couldn't put into a major acquisition because it was focused on the new product and Base44, so it bought back stock at a low price, with admittedly terrible short-term timing. Abrahami is unbothered: "The big question is where it's going to be in three years, not what happened in the last three months." He argues buybacks are a fantastic, underused tool, essentially a dividend to every shareholder, and companies should lean on them to balance stock-based compensation instead of endlessly diluting. 10. Execution, not finance. A low stock price makes M&A currency less valuable, but Abrahami says that's not his real constraint. Base44 was a one-person company; Wix had to build an entire company around it, staffing it with people pulled from the core. "I don't know how to do another one of those at the same time and have the same quality." The bottleneck on the next acquisition is execution capacity, not the balance sheet. 11. Chosen to be here. The one thing money buys beyond food security is freedom, and the deepest form of that freedom is knowing you're here by choice. "I'm here because I've chosen to be here. Nobody made me." He could move to Costa Rica or dance carnival in Brazil, and choosing to stay and run a public company through a crashing stock is where he finds his power. Money also made him more impatient and a bit lazier, and more rational because he's no longer deciding from fear. 12. The random Wednesday. Resilience starts with accepting the storm will come, because we assume that if yesterday was easy tomorrow will be too, and reality doesn't move in gentle slopes. "The worst thing that happens is probably some random thing on some random Wednesday. It's not something you get a lot of warning for." His anchor, borrowed from Babylon 5, is that you get there when you get there and the weapons you have are the weapons you have, so the only real question is whether you're doing the best you can with what you control.
0:15

You Can't Vibe-Code Trust Avishai Abrahami, Co-Founder & CEO of Wix , interviewed by Harry Stebbings (kevin andres) Summary: Wix trades at a $2.8B market cap on $2.1B of revenue while the market ascribes roughly zero value to a business throwing off $400M a year in free cash flow. Wix CEO Avishai Abrahami's argument is that the market can't yet price what AI actually threatens: the moat is trust and business logic, and neither gets vibe-coded away. His response is to own the disruptor (Base44), train his own narrow models, and stay committed through a storm he insists always arrives on a random Wednesday. 1. Trust is the moat. The real value of Salesforce is trust: JP Morgan and huge banks let it hold all their customer data, and the CRM itself is a small part of that. "What other platform will JP Morgan trust for their customers' data? None." That trust took years to build and can't be reconstructed by an agent scraping a database, so the companies whose value lives in trust survive the SaaS apocalypse while the ones reduced to piping get commoditized. 2. The business-logic wall. "You're not going to vibe-code Shopify no matter how good you are. The business logic is too hard." Wix tested this directly: they asked a team of professional developers to build the operating logic for a single hairdresser in Base44, gave up after a week, brought in a stronger team, and still failed two weeks later. Complex operational software is far harder than a demo suggests, which is why the pizza shop and the hairdresser stay Wix customers rather than build their own stack. 3. Own the disruptor. Wix bought Base44, a one-person company, for $80M, and it now does over $150M in ARR, roughly double what they paid. Abrahami frames the future as three buckets: owners who never want to build, owners who vibe-code everything themselves, and a mix in the middle over the next five or six years. Rather than bet on which wins, Wix owns the tool customers would defect to, so a customer who switches platforms still switches to Wix. 4. Trading on someone else's news. "Today we are trading on other companies' news. We're not trading on Wix news. We're trading on what OpenAI or Anthropic or Google are saying." Base44 alone, valued on vibe-coding peer multiples, should be worth around $8B, which means the market assigns less than zero to Wix's core. Abrahami's response is to detach: he doesn't wake up checking whether the stock moved 20%, because the only thing he can influence is the business. 5. The narrow model. Wix fine-tuned and combined its own models and now matches top-tier frontier quality on Base44 tasks at far lower cost. The logic: they sit on a huge stream of training data from watching what users try and where they fail, so a model built for Base44 can skip what frontier models carry, like knowledge of Chinese poetry, and go deep on what someone means when they say "build me a task manager to tell my boyfriend where he's wrong." A narrow target is easier to hit than a frontier model, and Wix already runs a trained model on website generation that's faster, cheaper, and makes fewer errors, retrained weekly on a live feedback loop. 6. Quality before cost. When Harry cites Chamath's claim that open source runs 14-16x cheaper, Abrahami pushes back: that holds for small tasks, but for something as complex as Base44 the savings land at 5-10%, and his own model runs 1-30% cheaper than frontier, not the order of magnitude people assume. More to the point, this is the wrong time to chase cost: "20% more quality, 20% less cost, I'll go for the quality." It's a brand-new market that's just starting, and the job now is to make the product better. 7. The but is very big. "We all give too much credit for AI. It's amazing, it's incredible, it's super powerful, but the but is pretty big." He asked Claude to write a safety protocol and got six mandatory gates, then pushed back on each one and watched the model cave until only one survived, downgrading the rest from "must test" to "might want to look at later." We over-trust these systems, and that reflex, treating a Reddit post as equivalent to research published in Nature, is where the danger lives. 8. Customer support still breaks. Wix has 3,500 people and its single biggest department is customer support, serving 192 countries. They tried hard not to build their own AI support agent, tested many off-the-shelf products, and concluded flatly: "It doesn't work. We tried, we tried again, it didn't work." The gap between hyped AI support startups and what actually ships in production is the tell that the technology is earlier than the marketing, maybe five years from being different. 9. Buybacks as dividends. Wix had $1.5B sitting in the bank it couldn't put into a major acquisition because it was focused on the new product and Base44, so it bought back stock at a low price, with admittedly terrible short-term timing. Abrahami is unbothered: "The big question is where it's going to be in three years, not what happened in the last three months." He argues buybacks are a fantastic, underused tool, essentially a dividend to every shareholder, and companies should lean on them to balance stock-based compensation instead of endlessly diluting. 10. Execution, not finance. A low stock price makes M&A currency less valuable, but Abrahami says that's not his real constraint. Base44 was a one-person company; Wix had to build an entire company around it, staffing it with people pulled from the core. "I don't know how to do another one of those at the same time and have the same quality." The bottleneck on the next acquisition is execution capacity, not the balance sheet. 11. Chosen to be here. The one thing money buys beyond food security is freedom, and the deepest form of that freedom is knowing you're here by choice. "I'm here because I've chosen to be here. Nobody made me." He could move to Costa Rica or dance carnival in Brazil, and choosing to stay and run a public company through a crashing stock is where he finds his power. Money also made him more impatient and a bit lazier, and more rational because he's no longer deciding from fear. 12. The random Wednesday. Resilience starts with accepting the storm will come, because we assume that if yesterday was easy tomorrow will be too, and reality doesn't move in gentle slopes. "The worst thing that happens is probably some random thing on some random Wednesday. It's not something you get a lot of warning for." His anchor, borrowed from Babylon 5, is that you get there when you get there and the weapons you have are the weapons you have, so the only real question is whether you're doing the best you can with what you control.

Gokul Rajaram

22,485 次观看 • 10 天前