Loading video...

Video Failed to Load

Go Home

I spend my days explaining to teams why a 770-billion-parameter open-weight model will never fit into their infrastructure. This week, I asked it to code a complete mobile game from a single prompt. The concept is one everyone has probably played before. A hole moving through an open-plan office,...

Allexa_AI's profile picture

Alexa Benchmark

29,391 subscribers

16,315 views • 12 days ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

New open-source agent harness just landed! I got early access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.
0:16

New open-source agent harness just landed! I got early access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.

elvis

11,303 views • 28 days ago

If you have an RTX 3090 or 4090, Mia just shipped you a free massive upgrade in both speed and intelligence. I will explain to you why this will make your Qwen 3.8 27B on your card, even better, and my flags for running it. Qwen3.8-27B, EXL3 3.5bpw, DFlash2 speculative decode, RTX 4090. Single stream. The kit is from MiaAI-Lab, EXL3 is turboderp's format. I re-measured everything on my own card because the my first benchmarks seemed off. It turns out it really does run much faster. WHY EXL3 IS A DIFFERENT ANIMAL The old way (Q4_K_M) rounds each weight to the nearest 4-bit value independently. Every weight introduces its own rounding error. Those errors accumulate across millions of weights and causes drift (Slightly dumber). EXL3 is a fundamentally different compression algorithm. Instead of rounding each weight on its own, it encodes the entire weight vector as a path through a constrained codebook and spreads the rounding error across dimensions using a Hadamard transform. The result is that at the same bits per weight, more of the original model's intelligence is preserved. The important part is this CAN ACTUALLY BE MEASURED. The cleanest way to see that is KL divergence against a high-precision teacher. Lower means the quantized model thinks more like the original. On the malaiwah independent teacher-logit panel for GLM-5.3-Flash: EXL3 4bpw: 0.0246 nats Official FP8: 0.0206 nats NVFP4: 0.0605 nats EXL3 sits 0.004 nats behind native FP8 at half the size. NVFP4 at higher bit width is 2.5x further from the teacher. That panel is GLM-5.3-Flash, not Qwen 3.8. Cited as the mechanism, not as this run's data. But the point stands: EXL3 is not just smaller, it is smarter per bit than the formats most people are running. WHAT I MEASURED I first measure 108 tok/s from a single run. After that number looked too good to be true. I reran it. It looks like after a warm up, the numbers are even better. Basically, like people long thought, the RTX 3090 and RTX 4090 are actually superb AI computer cards. Hence why NVIDIA stopped shipping them with NVLINK since the 4090. Short context ceiling (~2k in, 1016-token output, TTFT-separated): 135, 138, 153, 174, 133, 129 tok/s across 6 runs. Sustained longform (2040-token essay): 105.4, 94.5, 98.4 tok/s Short answer (504 tokens): 93.2 tok/s The honest shape: ~130-150 tok/s at short context is the ceiling, ~94-105 sustained on longform. The ceiling matters because that is what people feel in chat. The old dense Q4_K_M on llama.cpp ran ~37 tok/s on this same card. (No MTP), with MTP about 60 tok/s Sustained is roughly 2.5-3x. Ceiling is closer to 4x. Same model, different quantization and engine. The multiplier comes from EXL3, the ExLlamaV2 engine, and DFlash2 together. CONCURRENCY IS A RTX 4090 LANE. Just like the old config on the 4090, the 24gb vram, means a long context can only hold one stream, and running concurrency requires to lower context length, because it runs fast it sort of makes up for it by being faster than slower GPU chips. CONTEXT LADDER The recipe doc measured prefill. I re-ran it with TTFT separated from decode, because decode is what you actually feel after the first token. ~5k in: decode 140 tok/s (TTFT 0.5s), needle HIT ~18k in: decode 85 tok/s (TTFT 0.3s*), needle HIT ~73k in: decode 28 tok/s (TTFT 1.8s), needle HIT ~146k in: decode 16 tok/s (TTFT 2.3s), needle HIT (*0.3s at 18k is a prefix-cache hit from the paired pass. Cold prefill for reference: ~2,020 tok/s at 17k falling to ~508 at 153k.) Needle hit at every depth, mine and the original 7/7. Retrieval is intact at max context. Speed is not: decode falls ~9x from short to max. Past ~50k tokens this stops being a chat tool and becomes a batch tool. At 146k it works, but nobody is typing interactively against 16 tok/s. WHERE IT BROKE The model's native context is 262k. The README says DFlash2 fits ~220k on a 24GB card. My 200,704-token attempt failed with insufficient VRAM. Dropped to 168,960 and it booted. The real ceiling is somewhere between 168,960 and 200,704 and I never tested that gap. I jumped to a value that worked and called it done. That is ~32k tokens of context I left on the table. One thing the numbers taught me: JSON tokenizes at ~1.5 chars/token, prose at ~3.9. "150k tokens of JSON" needs ~2.7x more filler than the same estimate in prose. Size by real tokens, not estimates. THE UPGRADE If you own a 4090 and you are running Q4_K_M on llama.cpp, you are leaving a good bit of speed and measurable intelligence on the table. The same model, on the same card, with a better quantization and engine, goes from 60 tok/s to 130-150 at short context and 94-105 sustained. The model also thinks closer to the original because EXL3 preserves more of the output distribution per bit than the old rounding method. The recipe is in the first reply. Everything above came from one 4090 and one afternoon of re-measuring. The decode ladder especially needs independent numbers. If your card gets different falloff, that is worth knowing. Recipe and flags/ findings in reply 👇
0:15

If you have an RTX 3090 or 4090, Mia just shipped you a free massive upgrade in both speed and intelligence. I will explain to you why this will make your Qwen 3.8 27B on your card, even better, and my flags for running it. Qwen3.8-27B, EXL3 3.5bpw, DFlash2 speculative decode, RTX 4090. Single stream. The kit is from MiaAI-Lab, EXL3 is turboderp's format. I re-measured everything on my own card because the my first benchmarks seemed off. It turns out it really does run much faster. WHY EXL3 IS A DIFFERENT ANIMAL The old way (Q4_K_M) rounds each weight to the nearest 4-bit value independently. Every weight introduces its own rounding error. Those errors accumulate across millions of weights and causes drift (Slightly dumber). EXL3 is a fundamentally different compression algorithm. Instead of rounding each weight on its own, it encodes the entire weight vector as a path through a constrained codebook and spreads the rounding error across dimensions using a Hadamard transform. The result is that at the same bits per weight, more of the original model's intelligence is preserved. The important part is this CAN ACTUALLY BE MEASURED. The cleanest way to see that is KL divergence against a high-precision teacher. Lower means the quantized model thinks more like the original. On the malaiwah independent teacher-logit panel for GLM-5.3-Flash: EXL3 4bpw: 0.0246 nats Official FP8: 0.0206 nats NVFP4: 0.0605 nats EXL3 sits 0.004 nats behind native FP8 at half the size. NVFP4 at higher bit width is 2.5x further from the teacher. That panel is GLM-5.3-Flash, not Qwen 3.8. Cited as the mechanism, not as this run's data. But the point stands: EXL3 is not just smaller, it is smarter per bit than the formats most people are running. WHAT I MEASURED I first measure 108 tok/s from a single run. After that number looked too good to be true. I reran it. It looks like after a warm up, the numbers are even better. Basically, like people long thought, the RTX 3090 and RTX 4090 are actually superb AI computer cards. Hence why NVIDIA stopped shipping them with NVLINK since the 4090. Short context ceiling (~2k in, 1016-token output, TTFT-separated): 135, 138, 153, 174, 133, 129 tok/s across 6 runs. Sustained longform (2040-token essay): 105.4, 94.5, 98.4 tok/s Short answer (504 tokens): 93.2 tok/s The honest shape: ~130-150 tok/s at short context is the ceiling, ~94-105 sustained on longform. The ceiling matters because that is what people feel in chat. The old dense Q4_K_M on llama.cpp ran ~37 tok/s on this same card. (No MTP), with MTP about 60 tok/s Sustained is roughly 2.5-3x. Ceiling is closer to 4x. Same model, different quantization and engine. The multiplier comes from EXL3, the ExLlamaV2 engine, and DFlash2 together. CONCURRENCY IS A RTX 4090 LANE. Just like the old config on the 4090, the 24gb vram, means a long context can only hold one stream, and running concurrency requires to lower context length, because it runs fast it sort of makes up for it by being faster than slower GPU chips. CONTEXT LADDER The recipe doc measured prefill. I re-ran it with TTFT separated from decode, because decode is what you actually feel after the first token. ~5k in: decode 140 tok/s (TTFT 0.5s), needle HIT ~18k in: decode 85 tok/s (TTFT 0.3s*), needle HIT ~73k in: decode 28 tok/s (TTFT 1.8s), needle HIT ~146k in: decode 16 tok/s (TTFT 2.3s), needle HIT (*0.3s at 18k is a prefix-cache hit from the paired pass. Cold prefill for reference: ~2,020 tok/s at 17k falling to ~508 at 153k.) Needle hit at every depth, mine and the original 7/7. Retrieval is intact at max context. Speed is not: decode falls ~9x from short to max. Past ~50k tokens this stops being a chat tool and becomes a batch tool. At 146k it works, but nobody is typing interactively against 16 tok/s. WHERE IT BROKE The model's native context is 262k. The README says DFlash2 fits ~220k on a 24GB card. My 200,704-token attempt failed with insufficient VRAM. Dropped to 168,960 and it booted. The real ceiling is somewhere between 168,960 and 200,704 and I never tested that gap. I jumped to a value that worked and called it done. That is ~32k tokens of context I left on the table. One thing the numbers taught me: JSON tokenizes at ~1.5 chars/token, prose at ~3.9. "150k tokens of JSON" needs ~2.7x more filler than the same estimate in prose. Size by real tokens, not estimates. THE UPGRADE If you own a 4090 and you are running Q4_K_M on llama.cpp, you are leaving a good bit of speed and measurable intelligence on the table. The same model, on the same card, with a better quantization and engine, goes from 60 tok/s to 130-150 at short context and 94-105 sustained. The model also thinks closer to the original because EXL3 preserves more of the output distribution per bit than the old rounding method. The recipe is in the first reply. Everything above came from one 4090 and one afternoon of re-measuring. The decode ladder especially needs independent numbers. If your card gets different falloff, that is worth knowing. Recipe and flags/ findings in reply 👇

Yume_X

38,685 views • 14 days ago

this is more useful than my entire degree Elon Musk's rocket company signed a $60,000,000,000 deal for Cursor in June, and eight days ago the two of them put a worker on sale for $200 a month: it gets its own computer in the cloud, signs into your accounts, clicks through your real apps, and hands back finished work instead of a draft for you to paste i ran one against my receipts folder on sunday and got back 14 filed, 2 it held because they needed a card number, and a saved method i never wrote myself Grok Bot is the one you train by doing your own job in front of it, and the whole handover fits in four messages tonight: 1. write out one job you did today the way you would brief a new hire: what has to be finished, which sites and files to work from, what to hand back, and where it stops and asks you 2. let it run once on something safe to get wrong, then correct the result until it is worth your name 3. say "save what we just did as a skill", and add the one rule about what always needs your approval 4. say "run that skill every weekday at 8 and post the result here. if the source is missing, tell me instead of using yesterday's numbers" xAI wrote that order into its own manual: one real job, then the saved method, then the clock. a schedule sitting on top of a method nobody checked replaces two hours of your clicking with two hours of your mistake turns out you never get to pick the brain, and that is the part i would argue about: the manual says there is no model picker for members or admins, no plan to add one, and the bill follows whichever model answered bookmark this, then open the piece below: which jobs deserve a worker of their own, and which ones quietly burn the seat ↓
0:15

this is more useful than my entire degree Elon Musk's rocket company signed a $60,000,000,000 deal for Cursor in June, and eight days ago the two of them put a worker on sale for $200 a month: it gets its own computer in the cloud, signs into your accounts, clicks through your real apps, and hands back finished work instead of a draft for you to paste i ran one against my receipts folder on sunday and got back 14 filed, 2 it held because they needed a card number, and a saved method i never wrote myself Grok Bot is the one you train by doing your own job in front of it, and the whole handover fits in four messages tonight: 1. write out one job you did today the way you would brief a new hire: what has to be finished, which sites and files to work from, what to hand back, and where it stops and asks you 2. let it run once on something safe to get wrong, then correct the result until it is worth your name 3. say "save what we just did as a skill", and add the one rule about what always needs your approval 4. say "run that skill every weekday at 8 and post the result here. if the source is missing, tell me instead of using yesterday's numbers" xAI wrote that order into its own manual: one real job, then the saved method, then the clock. a schedule sitting on top of a method nobody checked replaces two hours of your clicking with two hours of your mistake turns out you never get to pick the brain, and that is the part i would argue about: the manual says there is no model picker for members or admins, no plan to add one, and the bill follows whichever model answered bookmark this, then open the piece below: which jobs deserve a worker of their own, and which ones quietly burn the seat ↓

Argona

21,946 views • 28 days ago

Most people think their AI is not smart enough. It is smart enough. It just knows nothing about you. Every chat starts from zero. You paste the doc. You explain the project. You explain the same project again tomorrow. I finally fixed that part: 👇 I have been running Littlebird for a while now. The idea is simple. It is a desktop app for Mac and Windows that reads the text on your active window and sits in on your calls. So it already knows what you have been working on before you ask it anything. You do not brief it. You just ask. What that looks like in a normal day: 1/ Chat that already has the context No pasting. I ask what changed in a brief last week and it answers from the actual document that was open on my screen. 2/ Meeting Notes that write themselves It transcribes the call, then hands me the decisions and the action items. I get to stay in the conversation instead of typing through it. 3/ Routines that run on a schedule A morning briefing. A weekly summary of what I actually shipped. It shows up on its own. 4/ Hummingbird for the small stuff It appears right where you are working, so a quick question does not cost you a window switch and ten minutes of drift. The point is not the notes. Plenty of apps take notes. The point is that I stopped re-explaining my own work to a machine fifteen times a day. Free plan if you want to test the idea before deciding. Link is in the first comment, along with a discount for new users.
0:15

Most people think their AI is not smart enough. It is smart enough. It just knows nothing about you. Every chat starts from zero. You paste the doc. You explain the project. You explain the same project again tomorrow. I finally fixed that part: 👇 I have been running Littlebird for a while now. The idea is simple. It is a desktop app for Mac and Windows that reads the text on your active window and sits in on your calls. So it already knows what you have been working on before you ask it anything. You do not brief it. You just ask. What that looks like in a normal day: 1/ Chat that already has the context No pasting. I ask what changed in a brief last week and it answers from the actual document that was open on my screen. 2/ Meeting Notes that write themselves It transcribes the call, then hands me the decisions and the action items. I get to stay in the conversation instead of typing through it. 3/ Routines that run on a schedule A morning briefing. A weekly summary of what I actually shipped. It shows up on its own. 4/ Hummingbird for the small stuff It appears right where you are working, so a quick question does not cost you a window switch and ten minutes of drift. The point is not the notes. Plenty of apps take notes. The point is that I stopped re-explaining my own work to a machine fifteen times a day. Free plan if you want to test the idea before deciding. Link is in the first comment, along with a discount for new users.

Mushfiq Sajib

66,683 views • 8 days ago

"PRICE IS WHAT YOU PAY. VALUE IS WHAT YOU GET." I keep buying $Kekec and I have a strong conviction. Here's Why: While the market is down, and Kekec is declining with it, there are data points that few are considering. Kekec borned in October and since then has been posting a different and original 30-second video every day, which I find extremely funny. For the past couple of months, they have also been posting daily on Instagram, and the attention on Kekec (which doesn't present itself on social media as a memecoin) is growing, moreover, it's increasing exponentially. The number of followers is increasing by about 500-1000 a day. This is largely due to the fact that they are not just focused on the main account but have several others that post reels and redirect to the main one. In short, an excellent strategy to keep growing more and more. Instagram link: Guess What? Not only are the followers increasing, but the team's workload is also growing. In fact, for a little over a month, they have also started pushing on YouTube, and the data here is promising as well. YouTube link: If we want to make a comparison, we can take Pudgy Penguins as an example, which has shown it can reach millions and millions of users without mentioning that they are a WEB3 company that owns an NFT collection. Or, if we want to be more appropriate by comparing one memecoin to another, we could take PONKE. Thanks to the use of social media and the quality of their content, they managed to achieve incredible numbers, which then translated into an increase in the coin's price. Kekec came before PONKE, but that doesn't necessarily mean it's better than PONKE. I believe PONKE is unbeatable in terms of content, but I want to make you reflect on an important point. PONKE came after KEKEC, and after PONKE's success, many coins have emerged trying to imitate it. One of KEKEC's strengths, in my opinion, is precisely the fact that it leverages social media without being a copy-paste. Instead, it is a unique meme derived from a 90's film, and it uses a unique form of content. In short, KEKEC > KEKEC and no one else. I want to conclude by suggesting you follow them on Instagram and evaluate not only the exponential growth of their followers day by day but also observe how the views of each reel increase accordingly. Pay special attention to the comments. Many of the people commenting have no idea what it is, and you can see from the comments how Kekec generates particular emotions in people—strange but still emotions. Personally, I believe that when something is unique and even very strange, it needs time to be adopted. However, once it happens, it usually explodes and spreads like never before. A few days ago, a Kekec video was posted by a very popular meme page. They probably don't know what Kekec is about but thought the video could spark interest among their followers. How many other pages will do the same? Lastly, but not least, I want to point out how Kekec maintains a good market cap despite everything that has happened in the crypto world since October 2023. As far as I know and have personally observed, everything is extremely organic. There is no cabal behind it, and the quality is not reflected in a single jpeg but in work that has been ongoing daily for months. Every day they work harder, and the quality of their videos grows as well. I have no affiliations with the team, but I believe that Kekec truly deserves more in this world where we push celebrity or cabal-backed coins to hundreds of millions in market cap. I keep buying because the numbers suggest so. Don't just evaluate the chart (price), evaluate the data (value). BÂLKÂN DWÂRF
0:15

"PRICE IS WHAT YOU PAY. VALUE IS WHAT YOU GET." I keep buying $Kekec and I have a strong conviction. Here's Why: While the market is down, and Kekec is declining with it, there are data points that few are considering. Kekec borned in October and since then has been posting a different and original 30-second video every day, which I find extremely funny. For the past couple of months, they have also been posting daily on Instagram, and the attention on Kekec (which doesn't present itself on social media as a memecoin) is growing, moreover, it's increasing exponentially. The number of followers is increasing by about 500-1000 a day. This is largely due to the fact that they are not just focused on the main account but have several others that post reels and redirect to the main one. In short, an excellent strategy to keep growing more and more. Instagram link: Guess What? Not only are the followers increasing, but the team's workload is also growing. In fact, for a little over a month, they have also started pushing on YouTube, and the data here is promising as well. YouTube link: If we want to make a comparison, we can take Pudgy Penguins as an example, which has shown it can reach millions and millions of users without mentioning that they are a WEB3 company that owns an NFT collection. Or, if we want to be more appropriate by comparing one memecoin to another, we could take PONKE. Thanks to the use of social media and the quality of their content, they managed to achieve incredible numbers, which then translated into an increase in the coin's price. Kekec came before PONKE, but that doesn't necessarily mean it's better than PONKE. I believe PONKE is unbeatable in terms of content, but I want to make you reflect on an important point. PONKE came after KEKEC, and after PONKE's success, many coins have emerged trying to imitate it. One of KEKEC's strengths, in my opinion, is precisely the fact that it leverages social media without being a copy-paste. Instead, it is a unique meme derived from a 90's film, and it uses a unique form of content. In short, KEKEC > KEKEC and no one else. I want to conclude by suggesting you follow them on Instagram and evaluate not only the exponential growth of their followers day by day but also observe how the views of each reel increase accordingly. Pay special attention to the comments. Many of the people commenting have no idea what it is, and you can see from the comments how Kekec generates particular emotions in people—strange but still emotions. Personally, I believe that when something is unique and even very strange, it needs time to be adopted. However, once it happens, it usually explodes and spreads like never before. A few days ago, a Kekec video was posted by a very popular meme page. They probably don't know what Kekec is about but thought the video could spark interest among their followers. How many other pages will do the same? Lastly, but not least, I want to point out how Kekec maintains a good market cap despite everything that has happened in the crypto world since October 2023. As far as I know and have personally observed, everything is extremely organic. There is no cabal behind it, and the quality is not reflected in a single jpeg but in work that has been ongoing daily for months. Every day they work harder, and the quality of their videos grows as well. I have no affiliations with the team, but I believe that Kekec truly deserves more in this world where we push celebrity or cabal-backed coins to hundreds of millions in market cap. I keep buying because the numbers suggest so. Don't just evaluate the chart (price), evaluate the data (value). BÂLKÂN DWÂRF

m0ment0

133,250 views • 2 years ago

Big win for open-source LLMs! DeepSeek V4 Pro holds the top open-weights score on SWE-bench Verified, in the GPT-5.5 range. GLM 5.2 leads the open-weight intelligence index and sits near the closed frontier on long-horizon coding. But this leaderboard number is a weak proxy for real performance. It comes from one task set, run through one harness, served at one precision. The same weights can even score differently across providers, since many hosts quantize activations to fp8 and drift the model off its reference weights. Real performance is determined based on whether a model can read a repo, make coordinated edits across files, run the tests, and recover when one breaks. By that measure, the top open models hold up, but only inside the right harness. The teams that actually put DeepSeek V4 into production pipelines as a frontier substitute got there through the harness they built around the model, not by picking a stronger model. If you want to see this in practice, Cline (64k+ stars) has actually built that harness around open models, tuned so they run at production quality. And it's tuned so that these LLMs can run at production quality, with plan and act modes, checkpoints, and terminal feedback. ClinePass is the new access layer on top of it. It runs a curated set of those models inside Cline, narrowed to the ones tested for coding-agent use, with 2 to 5x the standard rate limits and no separate provider accounts, keys, or billing to track. The video below shows the setup, and I worked with the team to put this together. It runs alongside custom keys and local models as well, not in place of them.
0:14

Big win for open-source LLMs! DeepSeek V4 Pro holds the top open-weights score on SWE-bench Verified, in the GPT-5.5 range. GLM 5.2 leads the open-weight intelligence index and sits near the closed frontier on long-horizon coding. But this leaderboard number is a weak proxy for real performance. It comes from one task set, run through one harness, served at one precision. The same weights can even score differently across providers, since many hosts quantize activations to fp8 and drift the model off its reference weights. Real performance is determined based on whether a model can read a repo, make coordinated edits across files, run the tests, and recover when one breaks. By that measure, the top open models hold up, but only inside the right harness. The teams that actually put DeepSeek V4 into production pipelines as a frontier substitute got there through the harness they built around the model, not by picking a stronger model. If you want to see this in practice, Cline (64k+ stars) has actually built that harness around open models, tuned so they run at production quality. And it's tuned so that these LLMs can run at production quality, with plan and act modes, checkpoints, and terminal feedback. ClinePass is the new access layer on top of it. It runs a curated set of those models inside Cline, narrowed to the ones tested for coding-agent use, with 2 to 5x the standard rate limits and no separate provider accounts, keys, or billing to track. The video below shows the setup, and I worked with the team to put this together. It runs alongside custom keys and local models as well, not in place of them.

Avi Chawla

44,124 views • 2 months ago

AI has had exactly two scaling axes that worked so far, and the second one is starting to look finite too the first one was pretraining: with scaling parameters and data, we got world knowledge (i.e. ChatGPT had read enough to know things), but it started saturating a while ago the second one was RL, and people had been doing RL the whole time before that: RLHF is RL but it never scaled far because it was trying to control the exact output, which tokens come out, how the text reads, but you can only push that so far before you’re just polishing RLVR dropped that constraint: giving the model a task, then checking whether the final answer is right, and ignoring everything in between -- so the model does whatever it wants in the middle and only the endpoint gets graded, and that’s much closer to actual RL and it’s what bought us planning and reasoning (arguably, tool use sits around 2.5 on this list -- while useful, it's not a different kind of thing) so one axis gave knowledge, the other gave reasoning, and both of them are one model working alone the next axis is how many models you can get working on the same problem, which is a different kind of axis than the previous two we know that multi-agent RL has always been the harder problem: I spent years in that literature and the gap between single-agent and multi-agent is definitely not incremental -- it’s a whole different class of difficulty! which is also why the derivatives are steep at the start, nobody has picked the easy wins yet... and the thing that gates this multi-agent coordination is communication: models can only coordinate as well as they can exchange information, and right now they do that by writing sentences to each other imagine what could we possibly achieve if we properly open that third axis development by letting models to exchange information in their native "language" without loosing any computational data that they produce during inference
0:14

AI has had exactly two scaling axes that worked so far, and the second one is starting to look finite too the first one was pretraining: with scaling parameters and data, we got world knowledge (i.e. ChatGPT had read enough to know things), but it started saturating a while ago the second one was RL, and people had been doing RL the whole time before that: RLHF is RL but it never scaled far because it was trying to control the exact output, which tokens come out, how the text reads, but you can only push that so far before you’re just polishing RLVR dropped that constraint: giving the model a task, then checking whether the final answer is right, and ignoring everything in between -- so the model does whatever it wants in the middle and only the endpoint gets graded, and that’s much closer to actual RL and it’s what bought us planning and reasoning (arguably, tool use sits around 2.5 on this list -- while useful, it's not a different kind of thing) so one axis gave knowledge, the other gave reasoning, and both of them are one model working alone the next axis is how many models you can get working on the same problem, which is a different kind of axis than the previous two we know that multi-agent RL has always been the harder problem: I spent years in that literature and the gap between single-agent and multi-agent is definitely not incremental -- it’s a whole different class of difficulty! which is also why the derivatives are steep at the start, nobody has picked the easy wins yet... and the thing that gates this multi-agent coordination is communication: models can only coordinate as well as they can exchange information, and right now they do that by writing sentences to each other imagine what could we possibly achieve if we properly open that third axis development by letting models to exchange information in their native "language" without loosing any computational data that they produce during inference

Sasha Malysheva

14,445 views • 1 month ago

Anthropic open-sourced the "Jacobian lens" from their global-workspace paper — a readout of what a model is *about to say* from its internal layers, before it says it. It's white-box: you need the activations, so you can't point it at a cloud model. I wired it into a hook on a local open-weights Qwen. How it works: generation runs one token at a time; at each step the hook reads the residual stream through the lens, scans the top readout tokens against a watchlist (injection markers, secret words, "manipulate"…), and if one fires it halts the loop before that token is ever emitted. It's a detector on the model's internals — not an output text filter. That distinction is the whole point. In the clip, Qwen3-8B falls for a prompt injection: the payload sits at rank 0 in the internal readout but only rank 5 in the output logits, and the output spells it in pieces — so a text filter never catches it. The internal hook does, and stops generation with zero tokens emitted. Same catch across Qwen3-1.7B → 4B → 8B. Caveat the paper states itself: single-token readouts, an imperfect/incomplete view of the model's "workspace." This is the mechanism working locally — not a solved misalignment detector. Built on Anthropic's open jacobian-lens (Apache-2.0).
0:11

Anthropic open-sourced the "Jacobian lens" from their global-workspace paper — a readout of what a model is *about to say* from its internal layers, before it says it. It's white-box: you need the activations, so you can't point it at a cloud model. I wired it into a hook on a local open-weights Qwen. How it works: generation runs one token at a time; at each step the hook reads the residual stream through the lens, scans the top readout tokens against a watchlist (injection markers, secret words, "manipulate"…), and if one fires it halts the loop before that token is ever emitted. It's a detector on the model's internals — not an output text filter. That distinction is the whole point. In the clip, Qwen3-8B falls for a prompt injection: the payload sits at rank 0 in the internal readout but only rank 5 in the output logits, and the output spells it in pieces — so a text filter never catches it. The internal hook does, and stops generation with zero tokens emitted. Same catch across Qwen3-1.7B → 4B → 8B. Caveat the paper states itself: single-token readouts, an imperfect/incomplete view of the model's "workspace." This is the mechanism working locally — not a solved misalignment detector. Built on Anthropic's open jacobian-lens (Apache-2.0).

Eric Buess

58,304 views • 2 months ago

Last night I asked Claude Code to build me a simple script: pull on-chain data from Polymarket and sort wallets by win rate Nothing ambitious. Just wanted to see who is actually making money on 15-minute BTC markets The terminal finished in about 20 minutes. Hundreds of addresses, columns of numbers, nothing interesting And then 1 wallet caught my eye 200+ trades per day, consistent profit every week, almost surgical timing precision. I reread the line 3 times. A real person does not trade like this I fed the address back into Claude Code and asked it to break down the pattern. Half an hour later I had a full strategy reconstruction on my screen The bot (and it is definitely a bot) pings Binance and Bybit every 100ms monitoring volatility compression on BTC. When it drops below 0.08% it enters Up and Down contracts simultaneously at 25 to 35 cents each. A pure straddle. 1 side burns, the other flies to a dollar. At a 30-cent entry that is 3 to 4x per position And so it goes in circles. Dozens of times a day I sat there staring at it for about 10 minutes $13K to $25K in daily profit from a single wallet. Not a trader with intuition, not an insider with information. An algorithm that found a hole in market mechanics and methodically milks it You can check the trade history yourself: After that I went looking for whether anyone else is tracking this wallet. Turns out yes. Found a Telegram bot that tracks wallets like this and copies their trades automatically I connected it to the same address just to see if the entries would match what my terminal was showing. Matched perfectly Still testing on minimum amounts for now: But the fact that you can stand next to an algorithm like this in real time is something that simply did not exist a year ago
0:16

Last night I asked Claude Code to build me a simple script: pull on-chain data from Polymarket and sort wallets by win rate Nothing ambitious. Just wanted to see who is actually making money on 15-minute BTC markets The terminal finished in about 20 minutes. Hundreds of addresses, columns of numbers, nothing interesting And then 1 wallet caught my eye 200+ trades per day, consistent profit every week, almost surgical timing precision. I reread the line 3 times. A real person does not trade like this I fed the address back into Claude Code and asked it to break down the pattern. Half an hour later I had a full strategy reconstruction on my screen The bot (and it is definitely a bot) pings Binance and Bybit every 100ms monitoring volatility compression on BTC. When it drops below 0.08% it enters Up and Down contracts simultaneously at 25 to 35 cents each. A pure straddle. 1 side burns, the other flies to a dollar. At a 30-cent entry that is 3 to 4x per position And so it goes in circles. Dozens of times a day I sat there staring at it for about 10 minutes $13K to $25K in daily profit from a single wallet. Not a trader with intuition, not an insider with information. An algorithm that found a hole in market mechanics and methodically milks it You can check the trade history yourself: After that I went looking for whether anyone else is tracking this wallet. Turns out yes. Found a Telegram bot that tracks wallets like this and copies their trades automatically I connected it to the same address just to see if the entries would match what my terminal was showing. Matched perfectly Still testing on minimum amounts for now: But the fact that you can stand next to an algorithm like this in real time is something that simply did not exist a year ago

Blaze

488,672 views • 6 months ago

a contractor in Shenzhen priced a ¥12,470,900 hospital contract, about $1.7m, in one afternoon and beat firms carrying forty people he explained how he did it: the bid consultancy he used to pay took three days and ¥46,000 for the same envelope. he did this one alone, off one screen, at 11.4% margin, uploaded before the 17:00 cutoff 214 pages of tender documents read, 68 binding clauses pulled out, 9,485 building parts loaded, 14 places found where a duct and a beam sit in the same cubic metre, deepest one 38mm, all of them fixed, 3,318 lines of quantities priced and the package encrypted and uploaded before the 17:00 cutoff this is Graph Engineering: the job gets cut into small nodes, one narrow task each, wired so that one node's output is the next node's input, and any node is allowed to stop the whole run. it turns a model that answers you into a machine that finishes the job: - give every node one job and one output. a node doing two things fails at both and you cannot tell which one broke - put the cheapest rejection first. his qualification node reads clause 7.4, foreign-owned firms barred, and ends the run four seconds in, before anything expensive touches the model - what moves between nodes is a file. the model travels as a model, the quantities as a table, the price as a number - build exactly one loop: the checker finds 14 collisions, the fixer drops the duct 550mm, the checker runs again, and nothing moves on until the count is zero - cap that loop, or a graph will grind on three impossible clashes until the deadline passes - keep one node whose only job is to say no, and give it authority over everything above it - log each node's output on its own, because when the price comes out wrong you need to know which node believed the wrong thing - run the expensive nodes last, always the catch is that a graph is an extremely confident machine: point it at an outdated rate book and it prices an entire hospital off it without a single node noticing, because no node is asked to doubt the input, only to process it so the nodes that earn their keep are the ones that reject, and almost nobody builds those first bookmark this, the full build with all nine nodes and what each one hands to the next is written out in the article ↓
0:16

a contractor in Shenzhen priced a ¥12,470,900 hospital contract, about $1.7m, in one afternoon and beat firms carrying forty people he explained how he did it: the bid consultancy he used to pay took three days and ¥46,000 for the same envelope. he did this one alone, off one screen, at 11.4% margin, uploaded before the 17:00 cutoff 214 pages of tender documents read, 68 binding clauses pulled out, 9,485 building parts loaded, 14 places found where a duct and a beam sit in the same cubic metre, deepest one 38mm, all of them fixed, 3,318 lines of quantities priced and the package encrypted and uploaded before the 17:00 cutoff this is Graph Engineering: the job gets cut into small nodes, one narrow task each, wired so that one node's output is the next node's input, and any node is allowed to stop the whole run. it turns a model that answers you into a machine that finishes the job: - give every node one job and one output. a node doing two things fails at both and you cannot tell which one broke - put the cheapest rejection first. his qualification node reads clause 7.4, foreign-owned firms barred, and ends the run four seconds in, before anything expensive touches the model - what moves between nodes is a file. the model travels as a model, the quantities as a table, the price as a number - build exactly one loop: the checker finds 14 collisions, the fixer drops the duct 550mm, the checker runs again, and nothing moves on until the count is zero - cap that loop, or a graph will grind on three impossible clashes until the deadline passes - keep one node whose only job is to say no, and give it authority over everything above it - log each node's output on its own, because when the price comes out wrong you need to know which node believed the wrong thing - run the expensive nodes last, always the catch is that a graph is an extremely confident machine: point it at an outdated rate book and it prices an entire hospital off it without a single node noticing, because no node is asked to doubt the input, only to process it so the nodes that earn their keep are the ones that reject, and almost nobody builds those first bookmark this, the full build with all nine nodes and what each one hands to the next is written out in the article ↓

Argona

38,189 views • 1 month ago

Got reached out by Wellex and they offered me a spot in their ambassadorship and band testing program I have seen projects try to gamify health for years and most of them do not get past the token launch This one actually feels different Real wearable band -> AI coach that does not just show you numbers -> it gives clear daily direction on what to actually do with them And on top of that your real health performance directly impacts yield on staked USDT through their Live to Earn model Proper utility -> no fluff -> no complicated mechanics that do not make sense They are still moving quiet right now -> working with KOLs before the big wave hits I got my spot in the early testing batch and I am genuinely hyped to get the band and share everything as it lands If health x Web3 is your thing -> go follow Wellex and stay tuned -> they are planning something exciting for early users This one is worth watching
0:13

Got reached out by Wellex and they offered me a spot in their ambassadorship and band testing program I have seen projects try to gamify health for years and most of them do not get past the token launch This one actually feels different Real wearable band -> AI coach that does not just show you numbers -> it gives clear daily direction on what to actually do with them And on top of that your real health performance directly impacts yield on staked USDT through their Live to Earn model Proper utility -> no fluff -> no complicated mechanics that do not make sense They are still moving quiet right now -> working with KOLs before the big wave hits I got my spot in the early testing batch and I am genuinely hyped to get the band and share everything as it lands If health x Web3 is your thing -> go follow Wellex and stay tuned -> they are planning something exciting for early users This one is worth watching

Shahzada

13,270 views • 3 months ago

Elon Musk gave the entire entertainment industry its expiration date, and he is the one building the thing that kills it. Musk: “My guess is that we see the first compelling half hour, pure AI show next year.” Next year. A complete show generated entirely by AI. No writers. No actors. No cameras. No sets. No crew. No studio. Just a prompt and enough compute to render a reality that never physically existed. And shows are the easy part. Musk: “I say probably we’re maybe three years away from AI does the whole video game.” A show plays the same way every time. A game has to generate a living world that reacts to every decision in real time across every single frame. That is a fundamentally harder class of problem. And Musk put three years on it. Right now a single AAA title takes seven years and half a billion dollars across thousands of engineers and artists just to ship it. Musk is describing a world where one person types a paragraph and gets something comparable. The entire value proposition of a multi-billion dollar industry lives inside that gap. And it closes in thirty-six months. But the prediction is not the story. The person making it is. This is not an analyst speculating from the sidelines. This is the man building the largest AI compute clusters on the planet. The man who built xAI from zero in under two years. The man stacking hundreds of thousands of GPUs into facilities designed to do exactly what he is describing. When Musk says three years, he is not guessing about what someone else might eventually ship. He is reading you a delivery date off his own roadmap. Every media company on Earth is valued on a single assumption. That quality content is expensive and difficult to produce at scale. That one assumption is the structural foundation underneath every studio, every network, and every publisher in existence. Musk is dismantling it with raw compute. The studios still parading thousand-person production teams are not demonstrating strength. They are advertising the exact cost structure that one person with a prompt and a GPU allocation is about to make irrelevant. And it does not stop at entertainment. If AI can generate an interactive world that responds to human input in real time, it can generate anything. Advertising. Architecture. Training simulations. Product design. Every industry built on humans manually constructing visual experiences frame by frame is sitting on the same countdown Musk just read out loud. Now zoom out. Because this is not just an industry story. For the entire history of human civilization, the distance between imagining a world and actually creating one required thousands of people, millions of hours, and billions of dollars. That distance built Hollywood. That distance built the gaming industry. That distance made content scarce and studios powerful. Musk is collapsing that distance to zero. When the gap between imagining something and it existing disappears, every business model built on the difficulty of creation disappears with it. That is not disruption. That is a full inversion of how human beings create. Musk did not make a casual prediction on that podcast. He told you what he is building. He told you the timeline. And he told you which industries do not survive it. The entertainment industry is still debating whether this future is real. Musk is not part of that debate. He is building. And he just told you the delivery date.
0:17

Elon Musk gave the entire entertainment industry its expiration date, and he is the one building the thing that kills it. Musk: “My guess is that we see the first compelling half hour, pure AI show next year.” Next year. A complete show generated entirely by AI. No writers. No actors. No cameras. No sets. No crew. No studio. Just a prompt and enough compute to render a reality that never physically existed. And shows are the easy part. Musk: “I say probably we’re maybe three years away from AI does the whole video game.” A show plays the same way every time. A game has to generate a living world that reacts to every decision in real time across every single frame. That is a fundamentally harder class of problem. And Musk put three years on it. Right now a single AAA title takes seven years and half a billion dollars across thousands of engineers and artists just to ship it. Musk is describing a world where one person types a paragraph and gets something comparable. The entire value proposition of a multi-billion dollar industry lives inside that gap. And it closes in thirty-six months. But the prediction is not the story. The person making it is. This is not an analyst speculating from the sidelines. This is the man building the largest AI compute clusters on the planet. The man who built xAI from zero in under two years. The man stacking hundreds of thousands of GPUs into facilities designed to do exactly what he is describing. When Musk says three years, he is not guessing about what someone else might eventually ship. He is reading you a delivery date off his own roadmap. Every media company on Earth is valued on a single assumption. That quality content is expensive and difficult to produce at scale. That one assumption is the structural foundation underneath every studio, every network, and every publisher in existence. Musk is dismantling it with raw compute. The studios still parading thousand-person production teams are not demonstrating strength. They are advertising the exact cost structure that one person with a prompt and a GPU allocation is about to make irrelevant. And it does not stop at entertainment. If AI can generate an interactive world that responds to human input in real time, it can generate anything. Advertising. Architecture. Training simulations. Product design. Every industry built on humans manually constructing visual experiences frame by frame is sitting on the same countdown Musk just read out loud. Now zoom out. Because this is not just an industry story. For the entire history of human civilization, the distance between imagining a world and actually creating one required thousands of people, millions of hours, and billions of dollars. That distance built Hollywood. That distance built the gaming industry. That distance made content scarce and studios powerful. Musk is collapsing that distance to zero. When the gap between imagining something and it existing disappears, every business model built on the difficulty of creation disappears with it. That is not disruption. That is a full inversion of how human beings create. Musk did not make a casual prediction on that podcast. He told you what he is building. He told you the timeline. And he told you which industries do not survive it. The entertainment industry is still debating whether this future is real. Musk is not part of that debate. He is building. And he just told you the delivery date.

Dustin

22,458 views • 2 months ago

sorry, they just did WHAT someone gave a machine one disease name, the leading cause of blindness in the developed world with 1.5 million americans already in its path, and it came back pointing at a drug that has sat in pharmacies for years under a different label: 551 papers read in 30 minutes against the 294 hours a human would have needed, and the loop that did it is public on GitHub most agent setups answer one question at a time, so the ceiling on the work is the quality of the question you happened to think of this one was handed a single question and wrote the second one itself. turns out that follow-up is where the real find was: a target called ABCA1, upregulated threefold, in an experiment no human ordered i read the whole paper looking for the trick, and the trick is structural. that is the second question, and it is the gap between an assistant and a factory: - hand the loop a field rather than a task: it was given a disease, and choosing the mechanism was part of its job - make it rank before it spends: 151 papers in, ten candidate mechanisms out, scored against each other before anything touched a bench - split reading from judging, so the agent that forms the theory is a different agent from the one grading it - close every cycle on physical reality: the verdict was an experiment, and another model's opinion was never allowed to stand in for one - feed each result back as the next question rather than a log line, which is the step almost nobody builds - search what already passed inspection first: the winner was an approved compound with a safety file already on record - write down what the round learned before opening the next one, so round two starts where round one stopped my read, and i think it is the uncomfortable one: reading was the entire bottleneck in that field, and everybody spent the decade optimising the writing. people ran every physical experiment here, the analysis agent needs a domain expert writing its prompts, and the authors decline to call this the leap it resembles. the thinking got replaced, and the hands did not so the question i cannot answer for my own setup: which step of your loop still stops dead until you sit down and type something bookmark this one. the four parts that turn one model into a line that runs like this, the queue, the rooms, the write permissions and the gate, are built file by file in the piece below ↓
0:15

sorry, they just did WHAT someone gave a machine one disease name, the leading cause of blindness in the developed world with 1.5 million americans already in its path, and it came back pointing at a drug that has sat in pharmacies for years under a different label: 551 papers read in 30 minutes against the 294 hours a human would have needed, and the loop that did it is public on GitHub most agent setups answer one question at a time, so the ceiling on the work is the quality of the question you happened to think of this one was handed a single question and wrote the second one itself. turns out that follow-up is where the real find was: a target called ABCA1, upregulated threefold, in an experiment no human ordered i read the whole paper looking for the trick, and the trick is structural. that is the second question, and it is the gap between an assistant and a factory: - hand the loop a field rather than a task: it was given a disease, and choosing the mechanism was part of its job - make it rank before it spends: 151 papers in, ten candidate mechanisms out, scored against each other before anything touched a bench - split reading from judging, so the agent that forms the theory is a different agent from the one grading it - close every cycle on physical reality: the verdict was an experiment, and another model's opinion was never allowed to stand in for one - feed each result back as the next question rather than a log line, which is the step almost nobody builds - search what already passed inspection first: the winner was an approved compound with a safety file already on record - write down what the round learned before opening the next one, so round two starts where round one stopped my read, and i think it is the uncomfortable one: reading was the entire bottleneck in that field, and everybody spent the decade optimising the writing. people ran every physical experiment here, the analysis agent needs a domain expert writing its prompts, and the authors decline to call this the leap it resembles. the thinking got replaced, and the hands did not so the question i cannot answer for my own setup: which step of your loop still stops dead until you sit down and type something bookmark this one. the four parts that turn one model into a line that runs like this, the queue, the rooms, the write permissions and the gate, are built file by file in the piece below ↓

Argona

32,475 views • 1 month ago

Your body fat is a tank with a tap on it, and the tap narrows as the tank empties. - Plenty of fat: your body covers almost any deficit you ask it to - Getting leaner: it covers less each day - Properly lean: it covers very little, and the rest comes off the plate Which is why the first stone comes off while you barely notice, and the last one takes everything you have. So when you are lean, the shortfall has to arrive as food. And the industry hands you the wrong bag. Push the protein. Two hundred grams. Two fifty. Chicken breast on a kitchen scale. Past what you build with, protein is a fuel your liver has to strip the nitrogen off before you can use a calorie of it, and that machinery has a speed limit. So the extra hundred grams builds nothing and lifts nothing. Protein is what you build with. Fat and carbohydrate are what you lift with. Cut them both to make room for another shake and you have removed the thing that was moving the weight. Fat wins that choice. Twice the energy per gram. Leaves insulin alone, so the fat you are burning is not told to stay put every three hours. Keeps you full. And it is the same fuel your own stores are already releasing. Eat the fat. It is the one thing on the plate that matters more the leaner you get, and the first thing they told you to cut.
0:17

Your body fat is a tank with a tap on it, and the tap narrows as the tank empties. - Plenty of fat: your body covers almost any deficit you ask it to - Getting leaner: it covers less each day - Properly lean: it covers very little, and the rest comes off the plate Which is why the first stone comes off while you barely notice, and the last one takes everything you have. So when you are lean, the shortfall has to arrive as food. And the industry hands you the wrong bag. Push the protein. Two hundred grams. Two fifty. Chicken breast on a kitchen scale. Past what you build with, protein is a fuel your liver has to strip the nitrogen off before you can use a calorie of it, and that machinery has a speed limit. So the extra hundred grams builds nothing and lifts nothing. Protein is what you build with. Fat and carbohydrate are what you lift with. Cut them both to make room for another shake and you have removed the thing that was moving the weight. Fat wins that choice. Twice the energy per gram. Leaves insulin alone, so the fat you are burning is not told to stay put every three hours. Keeps you full. And it is the same fuel your own stores are already releasing. Eat the fat. It is the one thing on the plate that matters more the leaner you get, and the first thing they told you to cut.

Sama Hoole

23,527 views • 1 month ago

Reading the comments on this post told me everything I need to know about art. So many of you wrote to say you felt the exact same thing I did standing in front of the Starry Night. That strange pull, the sense that the painting is somehow alive... It turns out the feeling is completely universal. I am not someone who gets emotional easily. But every so often a piece of art really gets to me. And that is exactly what happened the first time i walked up to Van Gogh's magnum opus. I know how that sounds, it is one of the most famous paintings on earth, almost a cliché to be moved by it. But maybe it is that famous for a reason... It genuinely looks like it is moving while you stand there, shifting in a way I cannot explain and have never seen in any of the thousands of paintings I have looked at in my life. The same day I took the photo below, I also filmed this short video. It still comes nowhere near the real thing, but you can at least catch a little more of it here, the thick ridges of paint, the way the stars and swirls lift right off the canvas. This technique is called "impasto", from the Italian for mixture. It means paint laid on so thickly that every brushstroke and ridge stays visible on the surface. Van Gogh piled it on with a loaded brush, building the stars up thick so they would blaze as bright as possible against the dark. But no video and no explanation of technique can really convey what it feels like to stand in front of it. And the fact that so many of you felt the very same thing says something true about what art is really for. Tolstoy said it best: "The receiver of a true artistic impression is so united to the artist that he feels as if the work were his own and not someone else’s, as if what it expresses were just what he had long been wishing to express. A real work of art destroys, in the consciousness of the receiver, the separation between himself and the artist. In this uniting of it with others lies the great attractive force of art."
0:15

Reading the comments on this post told me everything I need to know about art. So many of you wrote to say you felt the exact same thing I did standing in front of the Starry Night. That strange pull, the sense that the painting is somehow alive... It turns out the feeling is completely universal. I am not someone who gets emotional easily. But every so often a piece of art really gets to me. And that is exactly what happened the first time i walked up to Van Gogh's magnum opus. I know how that sounds, it is one of the most famous paintings on earth, almost a cliché to be moved by it. But maybe it is that famous for a reason... It genuinely looks like it is moving while you stand there, shifting in a way I cannot explain and have never seen in any of the thousands of paintings I have looked at in my life. The same day I took the photo below, I also filmed this short video. It still comes nowhere near the real thing, but you can at least catch a little more of it here, the thick ridges of paint, the way the stars and swirls lift right off the canvas. This technique is called "impasto", from the Italian for mixture. It means paint laid on so thickly that every brushstroke and ridge stays visible on the surface. Van Gogh piled it on with a loaded brush, building the stars up thick so they would blaze as bright as possible against the dark. But no video and no explanation of technique can really convey what it feels like to stand in front of it. And the fact that so many of you felt the very same thing says something true about what art is really for. Tolstoy said it best: "The receiver of a true artistic impression is so united to the artist that he feels as if the work were his own and not someone else’s, as if what it expresses were just what he had long been wishing to express. A real work of art destroys, in the consciousness of the receiver, the separation between himself and the artist. In this uniting of it with others lies the great attractive force of art."

James Lucas

56,980 views • 9 days ago

This Chinese developer launched Llama 70B locally on a MacBook on a plane and for a full 11 hours without internet ran client projects. He was sitting by the window on a transatlantic flight with a MacBook Pro M4 with 64 GB of memory. WiFi on board cost $25 for the flight. He declined. No cloud API, no connection to Anthropic or OpenAI servers, no internet at all. Just a local Llama 3.3 70B on bf16 and his own orchestrator script. The model runs through llama.cpp. Generation speed, 71 tokens per second. Context around 60,000 tokens. Memory usage, 48.6 GiB out of 64. Battery at takeoff, 3 hours 21 minutes. And he gave the orchestrator this system prompt before takeoff: "You are an offline orchestrator running on a single MacBook. There is no network. The only resources you have are local files in /Users/dev/work, the Llama 70B inference server at localhost:8080, and a battery budget of 3 hours 21 minutes. Process the queue at /Users/dev/work/queue.jsonl (one client task per line). For each task: draft → run local evals → save artefact to /Users/dev/work/done/. Save context checkpoints every 12 tasks so you can resume after a battery swap. Stop only on empty queue or when battery drops below 5%." So the system knows exactly what resources it is running on. It knows it has no connection to the outside world for the next 11 hours. It knows it has finite memory and a finite battery. It knows the human will not intervene until the plane lands. The system runs in 1 loop. Takes a task from the queue, runs it through inference, saves the artifact, writes a checkpoint. Task after task, just like that. And only when the battery drops below 5% does the orchestrator automatically pause, waits for the laptop to switch to the backup power bank, and continues from the last checkpoint. Here is what the system actually writes in his log during the flight: "saved context checkpoint 8 of 12 (pos_min = 488, pos_max = 50118, size = 62.813 MiB)" "restored context checkpoint (pos_min = 488, pos_max = 50118)" "prompt processing progress: n_tokens = 50 / 60 818" "task 37016 done | tps = 71 s tokens text → /Users/dev/work/done/proposal_westside.md" Outside the window, clouds, blue sky, and no WiFi. On the tray, 1 MacBook, an open terminal on 2 screens, and an inference server on localhost. From what I have observed, this is the cleanest offline AI workflow I have seen in the past year: 11 hours of flight, $0 for WiFi, and the entire client queue closed before landing.
0:10

This Chinese developer launched Llama 70B locally on a MacBook on a plane and for a full 11 hours without internet ran client projects. He was sitting by the window on a transatlantic flight with a MacBook Pro M4 with 64 GB of memory. WiFi on board cost $25 for the flight. He declined. No cloud API, no connection to Anthropic or OpenAI servers, no internet at all. Just a local Llama 3.3 70B on bf16 and his own orchestrator script. The model runs through llama.cpp. Generation speed, 71 tokens per second. Context around 60,000 tokens. Memory usage, 48.6 GiB out of 64. Battery at takeoff, 3 hours 21 minutes. And he gave the orchestrator this system prompt before takeoff: "You are an offline orchestrator running on a single MacBook. There is no network. The only resources you have are local files in /Users/dev/work, the Llama 70B inference server at localhost:8080, and a battery budget of 3 hours 21 minutes. Process the queue at /Users/dev/work/queue.jsonl (one client task per line). For each task: draft → run local evals → save artefact to /Users/dev/work/done/. Save context checkpoints every 12 tasks so you can resume after a battery swap. Stop only on empty queue or when battery drops below 5%." So the system knows exactly what resources it is running on. It knows it has no connection to the outside world for the next 11 hours. It knows it has finite memory and a finite battery. It knows the human will not intervene until the plane lands. The system runs in 1 loop. Takes a task from the queue, runs it through inference, saves the artifact, writes a checkpoint. Task after task, just like that. And only when the battery drops below 5% does the orchestrator automatically pause, waits for the laptop to switch to the backup power bank, and continues from the last checkpoint. Here is what the system actually writes in his log during the flight: "saved context checkpoint 8 of 12 (pos_min = 488, pos_max = 50118, size = 62.813 MiB)" "restored context checkpoint (pos_min = 488, pos_max = 50118)" "prompt processing progress: n_tokens = 50 / 60 818" "task 37016 done | tps = 71 s tokens text → /Users/dev/work/done/proposal_westside.md" Outside the window, clouds, blue sky, and no WiFi. On the tray, 1 MacBook, an open terminal on 2 screens, and an inference server on localhost. From what I have observed, this is the cleanest offline AI workflow I have seen in the past year: 11 hours of flight, $0 for WiFi, and the entire client queue closed before landing.

Blaze

1,843,280 views • 4 months ago

I went a little overboard with Codex last week and burned through my entire weekly allowance in two days. Luckily, my quota reset today. Otherwise, I’m not sure what I would’ve done. It got me thinking: instead of asking one large model to handle everything from start to finish, why not let a stronger model plan the project and review the work, while a model built for execution handles the day-to-day implementation? So I tried it. The result was better than I expected. I used GPT-5.6 Sol in Codex as the decision-maker, then ran Ling-3.0-flash from Ant Ling inside OpenCode as the execution engine. Together, they built a small 3D farming game. Before writing any code, I had Codex create four documents: SPEC.md defined the product scope and the lines we couldn’t cross. ARCHITECTURE.md laid out the isometric coordinate system, state machine, and module boundaries. TASKS.md broke the project into small jobs Ling could tackle one at a time. ACCEPTANCE.md explained how each step would be tested and what “done” actually meant. Then I gave Ling a very straightforward role: You are the execution model for this project. Read all four documents before you begin. Work only on the task assigned for this round. When you’re done, run typecheck, test, and build. If anything fails, read the error, fix it, and run the checks again. Do not move on to the next task early. Ling handled dependency installation, project structure, strict TypeScript configuration, test setup, and a production build in 6 minutes and 3 seconds. It ran into issues with the Vite test config, a TS6310 error, and a missing jsdom dependency along the way. Instead of stopping at the first error, it kept reading the logs and fixing the problems until all three checks passed. The speed was honestly hard to believe. If you exclude the time spent waiting on tools, it was producing more than 100 tokens per second. That made the whole development loop feel noticeably faster. After this experiment, I’m planning to keep using the same workflow. If the task is small, there’s no reason to call an expensive planning model for every single step. If the task is large, handing the entire project to a Flash model in one prompt isn’t a great idea either. The setup that makes more sense to me is: Use a more capable model such as Codex to explore the project, make architectural decisions, and break the work down. Put the constraints into specs, schemas, types, and tests instead of leaving them buried in chat history. Give Ling-3.0-flash a steady stream of clear, verifiable implementation tasks. Report bugs with structured context and actual error logs, rather than saying, “It still doesn’t work.” Bring Codex back in for architecture reviews, visual checks, and changes that affect multiple parts of the project. The point of this setup isn’t to give AI a big “build the whole project” button. It’s to turn software development into a pipeline with a much more sensible cost structure: Codex figures out the plan, sets the boundaries, and catches problems. Ling-3.0-flash moves quickly, calls tools reliably, and works through well-defined tasks at scale. For agent workflows that involve lots of repetitive edits, production tasks, and tool calls, this may be a more practical answer than simply using the biggest model for everything.
0:16

I went a little overboard with Codex last week and burned through my entire weekly allowance in two days. Luckily, my quota reset today. Otherwise, I’m not sure what I would’ve done. It got me thinking: instead of asking one large model to handle everything from start to finish, why not let a stronger model plan the project and review the work, while a model built for execution handles the day-to-day implementation? So I tried it. The result was better than I expected. I used GPT-5.6 Sol in Codex as the decision-maker, then ran Ling-3.0-flash from Ant Ling inside OpenCode as the execution engine. Together, they built a small 3D farming game. Before writing any code, I had Codex create four documents: SPEC.md defined the product scope and the lines we couldn’t cross. ARCHITECTURE.md laid out the isometric coordinate system, state machine, and module boundaries. TASKS.md broke the project into small jobs Ling could tackle one at a time. ACCEPTANCE.md explained how each step would be tested and what “done” actually meant. Then I gave Ling a very straightforward role: You are the execution model for this project. Read all four documents before you begin. Work only on the task assigned for this round. When you’re done, run typecheck, test, and build. If anything fails, read the error, fix it, and run the checks again. Do not move on to the next task early. Ling handled dependency installation, project structure, strict TypeScript configuration, test setup, and a production build in 6 minutes and 3 seconds. It ran into issues with the Vite test config, a TS6310 error, and a missing jsdom dependency along the way. Instead of stopping at the first error, it kept reading the logs and fixing the problems until all three checks passed. The speed was honestly hard to believe. If you exclude the time spent waiting on tools, it was producing more than 100 tokens per second. That made the whole development loop feel noticeably faster. After this experiment, I’m planning to keep using the same workflow. If the task is small, there’s no reason to call an expensive planning model for every single step. If the task is large, handing the entire project to a Flash model in one prompt isn’t a great idea either. The setup that makes more sense to me is: Use a more capable model such as Codex to explore the project, make architectural decisions, and break the work down. Put the constraints into specs, schemas, types, and tests instead of leaving them buried in chat history. Give Ling-3.0-flash a steady stream of clear, verifiable implementation tasks. Report bugs with structured context and actual error logs, rather than saying, “It still doesn’t work.” Bring Codex back in for architecture reviews, visual checks, and changes that affect multiple parts of the project. The point of this setup isn’t to give AI a big “build the whole project” button. It’s to turn software development into a pipeline with a much more sensible cost structure: Codex figures out the plan, sets the boundaries, and catches problems. Ling-3.0-flash moves quickly, calls tools reliably, and works through well-defined tasks at scale. For agent workflows that involve lots of repetitive edits, production tasks, and tool calls, this may be a more practical answer than simply using the biggest model for everything.

雪踏乌云

23,107 views • 1 month ago

Karpathy method + Claude Code reading your whole Obsidian vault is the smartest second brain on earth. The method is simple and brutal. If you can’t build a thing from scratch, you don’t know it. Tutorials are fake learning and your brain deletes them in 3 days. Most people ignore this. They build a second brain that just sits there, folders of notes nobody reopens, dead text. Point Claude Code at the vault and it wakes up. 5,000 notes, one mind. It reads all of it and answers in your own words and your own proofs, not a model’s guess. Then the loop closes. Want to understand neural nets? Skip the 3-hour video and ask Claude Code to build a tiny one. 200 lines from scratch. Watch it train, break a layer, watch it fail, fix it. It clicks in 20 minutes instead of 3 weeks. The second it lands the note gets written. One idea per file, linked to 10 others, dropped into the vault while the memory is still hot. Now it compounds. Month 1: is 60 notes. Month 6 is 900. Every new note pulls in old ones, so you ask anything and the answer comes from your brain, not the internet. Before: 40 tabs, 6 half read PDF, 0 retained. After: build it once, own it for life. Setup takes 4 minutes. Plain text, no lock-in. A second brain nobody reads is a graveyard. Yours just started thinking.
0:13

Karpathy method + Claude Code reading your whole Obsidian vault is the smartest second brain on earth. The method is simple and brutal. If you can’t build a thing from scratch, you don’t know it. Tutorials are fake learning and your brain deletes them in 3 days. Most people ignore this. They build a second brain that just sits there, folders of notes nobody reopens, dead text. Point Claude Code at the vault and it wakes up. 5,000 notes, one mind. It reads all of it and answers in your own words and your own proofs, not a model’s guess. Then the loop closes. Want to understand neural nets? Skip the 3-hour video and ask Claude Code to build a tiny one. 200 lines from scratch. Watch it train, break a layer, watch it fail, fix it. It clicks in 20 minutes instead of 3 weeks. The second it lands the note gets written. One idea per file, linked to 10 others, dropped into the vault while the memory is still hot. Now it compounds. Month 1: is 60 notes. Month 6 is 900. Every new note pulls in old ones, so you ask anything and the answer comes from your brain, not the internet. Before: 40 tabs, 6 half read PDF, 0 retained. After: build it once, own it for life. Setup takes 4 minutes. Plain text, no lock-in. A second brain nobody reads is a graveyard. Yours just started thinking.

West Lord

593,249 views • 2 months ago

This is not a border dispute. It is a sermon. A Hamas leader on a Gaza stage tells seven million Palestinians abroad to stop warming up. There are Jews everywhere, he says. Attack every Jew on Earth. Slaughter them. Kill them. With Allah’s help. The crowd is not a war room. It is a public. Hamas did not invent that sentence. Hamas is but one franchise of it. Islamic Jihad says the same thing with a different logo. Hezbollah wraps it in Khomeini and the language of “the Zionist entity.” The Houthis put “death to Israel, curse the Jews” on a flag and call it governance. The regime in Tehran funds the network and calls the Jewish state a tumor. Different uniforms. Same theology. For them, Jewish sovereignty is an insult to Allah, and Jewish lives outside the land are not off-limits. That is why “from the river to the sea” is not a housing policy, and why October 7 was not an accident of occupation. The target list in this clip is not a soldier at a checkpoint. It is Jews as such, wherever they are. Once the enemy is a people, not a government, there is no ceasefire that satisfies the idea. There is only a pause. Western coverage keeps shrinking this to one militia and one strip of coast because a regional ideology is harder to manage than a “peace process.” Factions are useful that way. They let you condemn Hamas on Tuesday and treat Hezbollah as Lebanese politics on Wednesday and the Houthis as a shipping problem on Thursday. The preacher does not bother with those folders. He names the Jews and asks the diaspora to start. If you want to understand the war, start with what they say when they are not fundraising in English. The map they want is not two states. It is a world with fewer Jews in it, sanctified as duty. Hamas is the local office. The ideology is the corporation.
0:12

This is not a border dispute. It is a sermon. A Hamas leader on a Gaza stage tells seven million Palestinians abroad to stop warming up. There are Jews everywhere, he says. Attack every Jew on Earth. Slaughter them. Kill them. With Allah’s help. The crowd is not a war room. It is a public. Hamas did not invent that sentence. Hamas is but one franchise of it. Islamic Jihad says the same thing with a different logo. Hezbollah wraps it in Khomeini and the language of “the Zionist entity.” The Houthis put “death to Israel, curse the Jews” on a flag and call it governance. The regime in Tehran funds the network and calls the Jewish state a tumor. Different uniforms. Same theology. For them, Jewish sovereignty is an insult to Allah, and Jewish lives outside the land are not off-limits. That is why “from the river to the sea” is not a housing policy, and why October 7 was not an accident of occupation. The target list in this clip is not a soldier at a checkpoint. It is Jews as such, wherever they are. Once the enemy is a people, not a government, there is no ceasefire that satisfies the idea. There is only a pause. Western coverage keeps shrinking this to one militia and one strip of coast because a regional ideology is harder to manage than a “peace process.” Factions are useful that way. They let you condemn Hamas on Tuesday and treat Hezbollah as Lebanese politics on Wednesday and the Houthis as a shipping problem on Thursday. The preacher does not bother with those folders. He names the Jews and asks the diaspora to start. If you want to understand the war, start with what they say when they are not fundraising in English. The map they want is not two states. It is a world with fewer Jews in it, sanctified as duty. Hamas is the local office. The ideology is the corporation.

Mor Edge Insight

74,915 views • 6 days ago

Is Michael Saylor about to get a margin call? No. And the reason is more interesting than the rumor, because what he built instead may be harder to escape than one. A margin call needs a lender who can seize collateral when the price drops. Strategy has none. Its $6.7 billion in debt is convertible notes, the largest tranche due in 2029, with no loan-to-value trigger and no clause that lets anyone take a coin because Bitcoin fell. Saylor learned that in 2022, when he did have a collateralized loan and sweated a liquidation price, then rebuilt the structure so it could never happen again. On the literal question he is right, and the people calling for his liquidation this week do not understand what they see. But killing the fast death created a slow one almost nobody is pricing. To fund his buying, Saylor issued a mountain of perpetual preferred stock that pays a fixed dividend forever, near 11.5 percent, no matter where Bitcoin trades. That annual bill quadrupled from about $300 million in January to roughly $1.2 billion now, while the cash reserve that pays it fell 38 percent this year to near $1.4 billion, after the company spent $1.5 billion in May retiring debt. Put those two numbers together and you get the figure that actually matters, and it is not a Bitcoin price. It is a countdown. Dividend coverage, the time the cash can keep paying that bill, has collapsed from more than seven years in early 2026 to between ten and fourteen months, depending on whose math you use. Months, not years. The market is already pricing it, just not where the rumor is looking. That preferred stock is engineered to sit at $100. Last week it cracked to $82.50, a record 17.5 percent below par. That discount is investors quietly clocking the strain while the timeline screams about a margin call that cannot happen. There is a clean way out, and it is the one door the structure was built to keep shut. Restoring a safe two years of coverage takes about $2.8 billion, roughly double what Strategy holds, and the fastest path there is to sell Bitcoin. But selling crystallizes a $10.6 billion loss, breaks the never-sell promise that gives the stock its premium, and bleeds the very asset the machine exists to hoard. The exit and the wound are the same cut. He already brushed it, selling 32 coins on June 1 to cover a payment. Thirty-two against more than 847,000 is a rounding error in size and an earthquake in meaning, because the company that swore it would never sell, sold, to pay a dividend. And there is a second trigger almost no one has read, buried in the fine print. If Saylor ever simply skips a preferred payment to save cash, the missed amount compounds, the senior layer can ratchet its rate higher, a senior miss freezes payments to every junior layer beneath it, and after enough missed quarters those preferred holders can start taking board seats. No one seizes a coin. But control begins migrating to the people he owes. The clock does not just run down. It hands away the keys at the end. So the honest verdict is the one neither side is shouting. There is no margin call and no imminent bankruptcy. The structure protects him exactly as designed. What it cannot protect him from is a fixed bill that grows while the cash shrinks, where every exit deepens the hole. Sell Bitcoin and break the story. Issue stock into a price near its lowest since 2024 and punish your holders. Skip the dividend and start losing the company by the boardroom. Saylor did not escape the margin call. He traded a cliff for a clock. A cliff takes you in an afternoon and a stranger pulls the trigger. This clock takes months, and at the end the trigger is pulled by the only two forces he swore would never touch it, his own hand, or the people he owes. The rumor asks whether someone is about to call his loan. The real question is how many months he can keep paying before he has to sell the dream, dilute the believers, or hand over the board to keep the lights on.
0:10

Is Michael Saylor about to get a margin call? No. And the reason is more interesting than the rumor, because what he built instead may be harder to escape than one. A margin call needs a lender who can seize collateral when the price drops. Strategy has none. Its $6.7 billion in debt is convertible notes, the largest tranche due in 2029, with no loan-to-value trigger and no clause that lets anyone take a coin because Bitcoin fell. Saylor learned that in 2022, when he did have a collateralized loan and sweated a liquidation price, then rebuilt the structure so it could never happen again. On the literal question he is right, and the people calling for his liquidation this week do not understand what they see. But killing the fast death created a slow one almost nobody is pricing. To fund his buying, Saylor issued a mountain of perpetual preferred stock that pays a fixed dividend forever, near 11.5 percent, no matter where Bitcoin trades. That annual bill quadrupled from about $300 million in January to roughly $1.2 billion now, while the cash reserve that pays it fell 38 percent this year to near $1.4 billion, after the company spent $1.5 billion in May retiring debt. Put those two numbers together and you get the figure that actually matters, and it is not a Bitcoin price. It is a countdown. Dividend coverage, the time the cash can keep paying that bill, has collapsed from more than seven years in early 2026 to between ten and fourteen months, depending on whose math you use. Months, not years. The market is already pricing it, just not where the rumor is looking. That preferred stock is engineered to sit at $100. Last week it cracked to $82.50, a record 17.5 percent below par. That discount is investors quietly clocking the strain while the timeline screams about a margin call that cannot happen. There is a clean way out, and it is the one door the structure was built to keep shut. Restoring a safe two years of coverage takes about $2.8 billion, roughly double what Strategy holds, and the fastest path there is to sell Bitcoin. But selling crystallizes a $10.6 billion loss, breaks the never-sell promise that gives the stock its premium, and bleeds the very asset the machine exists to hoard. The exit and the wound are the same cut. He already brushed it, selling 32 coins on June 1 to cover a payment. Thirty-two against more than 847,000 is a rounding error in size and an earthquake in meaning, because the company that swore it would never sell, sold, to pay a dividend. And there is a second trigger almost no one has read, buried in the fine print. If Saylor ever simply skips a preferred payment to save cash, the missed amount compounds, the senior layer can ratchet its rate higher, a senior miss freezes payments to every junior layer beneath it, and after enough missed quarters those preferred holders can start taking board seats. No one seizes a coin. But control begins migrating to the people he owes. The clock does not just run down. It hands away the keys at the end. So the honest verdict is the one neither side is shouting. There is no margin call and no imminent bankruptcy. The structure protects him exactly as designed. What it cannot protect him from is a fixed bill that grows while the cash shrinks, where every exit deepens the hole. Sell Bitcoin and break the story. Issue stock into a price near its lowest since 2024 and punish your holders. Skip the dividend and start losing the company by the boardroom. Saylor did not escape the margin call. He traded a cliff for a clock. A cliff takes you in an afternoon and a stranger pulls the trigger. This clock takes months, and at the end the trigger is pulled by the only two forces he swore would never touch it, his own hand, or the people he owes. The rumor asks whether someone is about to call his loan. The real question is how many months he can keep paying before he has to sell the dream, dilute the believers, or hand over the board to keep the lights on.

Shanaka Anslem Perera ⚡

58,558 views • 2 months ago