Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

VisionPsy-Nano on an iPhone 15. Same chart, same question, four models racing. 0.9s. Every other model took 2-13x longer and still got the answer wrong.

0xMarioNawfal

258,565 subscribers

65,059 views • 1 month ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Not every task needs the same model. A quick summary doesn't need the same horsepower as a deep research question — and it shouldn't cost the same either. Now, you can now easily compare models for your Custom Agent on speed, intelligence, and cost 🫡

Not every task needs the same model. A quick summary doesn't need the same horsepower as a deep research question — and it shouldn't cost the same either. Now, you can now easily compare models for your Custom Agent on speed, intelligence, and cost 🫡

Notion

32,420 views • 6 months ago

So I got my 2nd replacement for my iPhone 15 Pro Max from Apple and it still has the same top speaker issue?!?! When the volume is above 80% with some songs and videos the speaker rattles and cracks like it has water in it. I tried this at BestBuy too and they had the same issue

So I got my 2nd replacement for my iPhone 15 Pro Max from Apple and it still has the same top speaker issue?!?! When the volume is above 80% with some songs and videos the speaker rattles and cracks like it has water in it. I tried this at BestBuy too and they had the same issue

MilesAboveTech

702,281 views • 2 years ago

The fifth episode of #NowBingeing is LIVE! Watch the latest episode with Vijay Varma and answer this simple question! Follow these simple steps: 🌟 Comment your answer to the question above, and share the post with the answer on your story! 🌟Use the #NowBingeing and tag 2 friends to do the same 🌟Lucky winners will stand a chance to win exciting vouchers!

The fifth episode of #NowBingeing is LIVE! Watch the latest episode with Vijay Varma and answer this simple question! Follow these simple steps: 🌟 Comment your answer to the question above, and share the post with the answer on your story! 🌟Use the #NowBingeing and tag 2 friends to do the same 🌟Lucky winners will stand a chance to win exciting vouchers!

Tata Play Binge

28,068 views • 3 years ago

🚨 JANE STREET STRIKES $BTC AGAIN It started Monday. Four sessions, $51 billion in market cap erased. No headline. No warning. Four days. Four US opens. Four dumps. August 11: opens, drops 1.7% August 12: opens, drops 1.7% August 13: opens, drops 1.4% August 14: opens, drops 1.9% Every night Bitcoin climbs. Every US open erases it. $65,400 to $62,800, and every dollar went in the same window. Today was the tell. PPI came in soft. Bitcoin ripped to $64,000. New York opened. Straight to $62,800. Good news at 8:30. Sold by 10:00. Four for four is a 6% coincidence if you believe in coincidences. Same hour. Same size. Same silence around it. Bookmark this and check the open on Monday. If it prints a fifth time, you will never look at a chart the same way again. I post the moment this pattern breaks. Follow and turn notifications on.

🚨 JANE STREET STRIKES $BTC AGAIN It started Monday. Four sessions, $51 billion in market cap erased. No headline. No warning. Four days. Four US opens. Four dumps. August 11: opens, drops 1.7% August 12: opens, drops 1.7% August 13: opens, drops 1.4% August 14: opens, drops 1.9% Every night Bitcoin climbs. Every US open erases it. $65,400 to $62,800, and every dollar went in the same window. Today was the tell. PPI came in soft. Bitcoin ripped to $64,000. New York opened. Straight to $62,800. Good news at 8:30. Sold by 10:00. Four for four is a 6% coincidence if you believe in coincidences. Same hour. Same size. Same silence around it. Bookmark this and check the open on Monday. If it prints a fifth time, you will never look at a chart the same way again. I post the moment this pattern breaks. Follow and turn notifications on.

Nonzee

17,475 views • 22 days ago

THE CROWD HAS BEEN WRONG AT EVERY MAJOR TURN THIS TIME IS NO DIFFERENT In 2022 they said $20K was the bottom it wasn't In 2023 they said crypto was dead it wasn't In 2025 they said $126K was just the beginning it wasn't Right now the crowd is convinced the worst is behind us They're wrong again The same sentiment, the same narratives, the same misplaced confidence that shows up at every major top and before every major flush Below $50K is coming not because of doom and gloom, because the chart, the cycle, the math, and the liquidity map all point to the same place The crowd will understand it after See you below $50K Turn notifications on - I'll call it before it prints

THE CROWD HAS BEEN WRONG AT EVERY MAJOR TURN THIS TIME IS NO DIFFERENT In 2022 they said $20K was the bottom it wasn't In 2023 they said crypto was dead it wasn't In 2025 they said $126K was just the beginning it wasn't Right now the crowd is convinced the worst is behind us They're wrong again The same sentiment, the same narratives, the same misplaced confidence that shows up at every major top and before every major flush Below $50K is coming not because of doom and gloom, because the chart, the cycle, the math, and the liquidity map all point to the same place The crowd will understand it after See you below $50K Turn notifications on - I'll call it before it prints

ardizor 🧙‍♂️

48,461 views • 2 months ago

Kimi K3 beats GPT 5.6 Sol on both speed and in cost in backend bug fixing ! Planted 4 defects in the same Python order service: a wrong-total calculation, a broken access-control check (IDOR), a crash on invalid input, and an inventory oversell. Both models got identical code and the same prompt, with an automated test suite verifying every fix over the live API. Both went 4/4, but by the time Kimi K3 finished all 4, GPT 5.6 had only fixed 2. Kimi wrapped in under half the time, at half the cost. Kimi K3: 1min 34s · $0.1 · 4/4 GPT 5.6 Sol: 4min 6s · $0.2 · 4/4

Kimi K3 beats GPT 5.6 Sol on both speed and in cost in backend bug fixing ! Planted 4 defects in the same Python order service: a wrong-total calculation, a broken access-control check (IDOR), a crash on invalid input, and an inventory oversell. Both models got identical code and the same prompt, with an automated test suite verifying every fix over the live API. Both went 4/4, but by the time Kimi K3 finished all 4, GPT 5.6 had only fixed 2. Kimi wrapped in under half the time, at half the cost. Kimi K3: 1min 34s · $0.1 · 4/4 GPT 5.6 Sol: 4min 6s · $0.2 · 4/4

GMI Cloud

24,671 views • 1 month ago

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.

Sudo su

72,554 views • 6 months ago

Meta paid $250M for ONE AI researcher Lightricks built an entire frontier model for less than that.. LTX-2 now #3 globally while tech giants burn billions on the same results craft beats cash every time Here's why:

Meta paid $250M for ONE AI researcher Lightricks built an entire frontier model for less than that.. LTX-2 now #3 globally while tech giants burn billions on the same results craft beats cash every time Here's why:

Robin Delta

26,882 views • 9 months ago

Pakistani Journalists : "In IPL, A 15 Year old Vaibhav Sooryavanshi Scored 36 Balls 100 today and on the same day you scored 40 balls 50, Don't you think PAKISTAN need to change their game? Babar Azam :- "Leave it, I don't want to answer your question"

Pakistani Journalists : "In IPL, A 15 Year old Vaibhav Sooryavanshi Scored 36 Balls 100 today and on the same day you scored 40 balls 50, Don't you think PAKISTAN need to change their game? Babar Azam :- "Leave it, I don't want to answer your question"

Rohit

209,911 views • 4 months ago

I STARED AT THIS FOR TEN MINUTES TRYING TO PROVE IT WRONG. I COULDN'T. I overlaid Google on Bitcoin. Google broke its 2021 high, retested it, then went vertical to $350. Bitcoin broke its 2021 high. And it's retesting the exact same zone right now. One is a $2T tech giant. The other is Bitcoin. Same chart. One cycle apart. You already know how the Google one ended.

I STARED AT THIS FOR TEN MINUTES TRYING TO PROVE IT WRONG. I COULDN'T. I overlaid Google on Bitcoin. Google broke its 2021 high, retested it, then went vertical to $350. Bitcoin broke its 2021 high. And it's retesting the exact same zone right now. One is a $2T tech giant. The other is Bitcoin. Same chart. One cycle apart. You already know how the Google one ended.

Merlijn The Trader

90,815 views • 1 month ago

AI that watches your work! 👀 Codya built AI that watches assembly lines and verifies every step got done right. Engine being assembled, bolt by bolt. The model tracks each one, inserted vs fastened. Standard computer vision would struggle here. It's about occlusion. A worker's hands constantly cover the bolts during assembly. Hand moves in, bolt disappears. Hand moves away, bolt reappears. Without smart tracking, the model forgets which bolt is which every time it's hidden. They solved it with BoT-SORT, a tracking algorithm that keeps object identity through occlusions. When a hand covers a bolt and moves away, the tracker knows it's still the same bolt, same state, same position. SkalskiP looking forward to see more and more industrial use cases with your Roboflow models :) ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

AI that watches your work! 👀 Codya built AI that watches assembly lines and verifies every step got done right. Engine being assembled, bolt by bolt. The model tracks each one, inserted vs fastened. Standard computer vision would struggle here. It's about occlusion. A worker's hands constantly cover the bolts during assembly. Hand moves in, bolt disappears. Hand moves away, bolt reappears. Without smart tracking, the model forgets which bolt is which every time it's hidden. They solved it with BoT-SORT, a tracking algorithm that keeps object identity through occlusions. When a hand covers a bolt and moves away, the tracker knows it's still the same bolt, same state, same position. SkalskiP looking forward to see more and more industrial use cases with your Roboflow models :) ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

39,467 views • 1 month ago

yesterday it was Windows XP. today, it’s DOOM on an ESP32. a tiny microcontroller, a microSD card, and a round LCD wired up on a breadboard - and the classic game is running on the display. the funny part isn’t just that DOOM keeps appearing on unexpected hardware. it’s that every new platform eventually gets asked the same question: “can it run DOOM?” even an ESP32.

yesterday it was Windows XP. today, it’s DOOM on an ESP32. a tiny microcontroller, a microSD card, and a round LCD wired up on a breadboard - and the classic game is running on the display. the funny part isn’t just that DOOM keeps appearing on unexpected hardware. it’s that every new platform eventually gets asked the same question: “can it run DOOM?” even an ESP32.

ard

134,533 views • 1 month ago

CHATGPT STILL CAN’T BUILD SITES THAT LOOK LIKE THIS Only Claude is building websites that actually feel expensive right now. Scroll through the clip. Particle systems. Liquid transitions. Cinematic depth. Every frame feels like a high-budget trailer. Claude produces this level of motion and atmosphere. ChatGPT still outputs flat layouts and basic interactions that look dated the moment they load. The gap is no longer small. One model delivers premium digital experiences. The other still struggles to clear “clean and modern.” Agencies and freelancers who switched already charge more and deliver faster. The ones still prompting the wrong model keep losing the high-ticket clients.

CHATGPT STILL CAN’T BUILD SITES THAT LOOK LIKE THIS Only Claude is building websites that actually feel expensive right now. Scroll through the clip. Particle systems. Liquid transitions. Cinematic depth. Every frame feels like a high-budget trailer. Claude produces this level of motion and atmosphere. ChatGPT still outputs flat layouts and basic interactions that look dated the moment they load. The gap is no longer small. One model delivers premium digital experiences. The other still struggles to clear “clean and modern.” Agencies and freelancers who switched already charge more and deliver faster. The ones still prompting the wrong model keep losing the high-ticket clients.

Romario

94,743 views • 28 days ago

You can now use GPT 5.5, Gemini 3.7 Flash, Kimi K3 and 47 other AI models completely free😱 No subscription. No credit card. Even the API usage costs $0. AIHubMix just opened a free catalog with 50 AI models. Some of the available models: • Ox Alpha • Gemini 3.7 Flash • GLM 5.2 • Kimi K3 • MiniMax M3 • GPT 5.5 • 40+ more And you don’t need separate API keys for each model. Setup takes 2 minutes: > Step 1: Go to > Create an account using your email or OAuth. No card needed. Step 2: Create one API key > The same key works with every free and paid model. Step 3: Add it to any OpenAI-compatible tool Base URL: Then choose any model ending in -free, such as: coding-glm-5.2-free gpt-5.5-free That’s it. One API key. 50 AI models. $0 for both input and output. Save this. You might need a free multi-model setup later.

You can now use GPT 5.5, Gemini 3.7 Flash, Kimi K3 and 47 other AI models completely free😱 No subscription. No credit card. Even the API usage costs $0. AIHubMix just opened a free catalog with 50 AI models. Some of the available models: • Ox Alpha • Gemini 3.7 Flash • GLM 5.2 • Kimi K3 • MiniMax M3 • GPT 5.5 • 40+ more And you don’t need separate API keys for each model. Setup takes 2 minutes: > Step 1: Go to > Create an account using your email or OAuth. No card needed. Step 2: Create one API key > The same key works with every free and paid model. Step 3: Add it to any OpenAI-compatible tool Base URL: Then choose any model ending in -free, such as: coding-glm-5.2-free gpt-5.5-free That’s it. One API key. 50 AI models. $0 for both input and output. Save this. You might need a free multi-model setup later.

CDG

15,278 views • 13 days ago

Got a 16GB GPU? You can run all of these right now. Tested 4 Qwen3.5-based models on ToolCall-15 & BugFind-15: Models: - Qwen3.5:9b Q8 (Official) - Qwopus v3 Q8 by Jackrong - OmniCoder-9B by Tesslate - Qwen3.5-9b-Sushi-Coder by bigatuna Summary: - ToolCall-15: Qwopus v3 went perfect 30/30, Sushicoder beat base Qwen3.5 - BugFind-15: Omnicoder flipped the script and took #1 at 83% No single model won both, that's the fun part. Open source community is cooking.

Got a 16GB GPU? You can run all of these right now. Tested 4 Qwen3.5-based models on ToolCall-15 & BugFind-15: Models: - Qwen3.5:9b Q8 (Official) - Qwopus v3 Q8 by Jackrong - OmniCoder-9B by Tesslate - Qwen3.5-9b-Sushi-Coder by bigatuna Summary: - ToolCall-15: Qwopus v3 went perfect 30/30, Sushicoder beat base Qwen3.5 - BugFind-15: Omnicoder flipped the script and took #1 at 83% No single model won both, that's the fun part. Open source community is cooking.

stevibe

75,514 views • 5 months ago

My first test with Mimimax H3 - Early Access One character. Four careers. Zero breaks in consistency. Built this cinematic transformation with MiniMax H3 on MiniMax Design (H3) using Omni Reference, keeping the same character perfectly consistent as she seamlessly transforms from a CEO to a runway model, an astronaut, and finally a professional racing driver. The character consistency, seamless outfit changes, and cinematic transitions feel incredibly polished more like a film sequence than a typical AI video. #MiniMaxH3

My first test with Mimimax H3 - Early Access One character. Four careers. Zero breaks in consistency. Built this cinematic transformation with MiniMax H3 on MiniMax Design (H3) using Omni Reference, keeping the same character perfectly consistent as she seamlessly transforms from a CEO to a runway model, an astronaut, and finally a professional racing driver. The character consistency, seamless outfit changes, and cinematic transitions feel incredibly polished more like a film sequence than a typical AI video. #MiniMaxH3

Smiling Khan

16,657 views • 1 month ago

Four members of the same family were killed and three other persons were left injured after an SUV crashed into the family's car in a tragic accident near Saharanpur on Friday, officials said. The incident occurred on the Dehradun-Panchkula Highway within the Rampur Maniharan area of Saharanpur when the speeding Bolero rammed the rear of the Tata Tiago car.

Four members of the same family were killed and three other persons were left injured after an SUV crashed into the family's car in a tragic accident near Saharanpur on Friday, officials said. The incident occurred on the Dehradun-Panchkula Highway within the Rampur Maniharan area of Saharanpur when the speeding Bolero rammed the rear of the Tata Tiago car.

NDTV

1,272,037 views • 2 months ago

Yup, a football video. The World Cup made us do it Luma rebuilt image generation from scratch — reasoning first, pixels second. And it beats Google's Nano Banana 2 and GPT Image 1.5 on reasoning benchmarks All 3 new models are now live on AI/ML API luma/uni-1 plans before it draws. The model generates autoregressively: it works out layout, composition and text placement first, then renders the pixels. $0.052/image luma/uni-1-max — same prompts, same params, max fidelity. 2K output + editing with up to 9 reference images. Built for hero shots and ad creative. $0.13/image luma/ray-3-2 — up to 16 keyframes per clip, 20s, 1080p, native HDR + 16-bit EXR export. The video in this post came straight out of it model ids "luma/uni-1" "luma/uni-1-max" "luma/ray-3-2" Luma cooked. We serve

Yup, a football video. The World Cup made us do it Luma rebuilt image generation from scratch — reasoning first, pixels second. And it beats Google's Nano Banana 2 and GPT Image 1.5 on reasoning benchmarks All 3 new models are now live on AI/ML API luma/uni-1 plans before it draws. The model generates autoregressively: it works out layout, composition and text placement first, then renders the pixels. $0.052/image luma/uni-1-max — same prompts, same params, max fidelity. 2K output + editing with up to 9 reference images. Built for hero shots and ad creative. $0.13/image luma/ray-3-2 — up to 16 keyframes per clip, 20s, 1080p, native HDR + 16-bit EXR export. The video in this post came straight out of it model ids "luma/uni-1" "luma/uni-1-max" "luma/ray-3-2" Luma cooked. We serve

AI/ML API

19,375 views • 1 month ago

🔥 Shedeur. Focused on the Bills 💛 "He's got the same attitude every single day. You have to be able to turn the page after a loss, after win. You can't take the previous game with you and I think he does an excellent job of that"

🔥 Shedeur. Focused on the Bills 💛 "He's got the same attitude every single day. You have to be able to turn the page after a loss, after win. You can't take the previous game with you and I think he does an excellent job of that"

JaKi 🇺🇸

31,756 views • 8 months ago

I built an AI fitness influencer on Instagram She posts every day. Never shows up late. Made me $1,400 last month Full stack: - Nano Banana 2 → character consistency - Kling 3.0 → video generation - ElevenLabs → voiceover - Claude → captions + hooks - CapCut → final cut Cost: ~$180/mo Revenue: $1,200/mo Time: 20 min/day Same niche as real creators. 1/50th of the work

I built an AI fitness influencer on Instagram She posts every day. Never shows up late. Made me $1,400 last month Full stack: - Nano Banana 2 → character consistency - Kling 3.0 → video generation - ElevenLabs → voiceover - Claude → captions + hooks - CapCut → final cut Cost: ~$180/mo Revenue: $1,200/mo Time: 20 min/day Same niche as real creators. 1/50th of the work

SoloTech

203,033 views • 4 months ago