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.

Codex GPT-5.6 Sol + Seedance 2.0 is the way to create such scrolly-telling websites GitHub open-source: Inspo: Peter Wang Co-work: AI Pulse Save & re-use it forever

AmirMušić

74,801 subscribers

54,387 views • 16 days ago •via X (Twitter)

Science & Technology Education

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

You can now orchestrate Fable 5, Sol, and any model inside Codex with one plugin. It's called Codex-Orchestration. Assign Fable 5 as the advisor, Sol as the executor, or any model to any role. Then define the order they work in. Codex handles the routing. I ran Fable 5 High as planner with GPT-5.6 Sol Extra High as executor on a set of issues Opus and GPT-5.5 always struggled with. Done in 30 minutes. 40% fewer limit hits. 2x faster implementation. Install it by pasting this into Codex: "Install Codex Orchestration: codex plugin marketplace add Cjbuilds/Codex-Orchestration codex plugin add codex-orchestration@codex-orchestration Verify the installation, then tell me to start a new task." Then assign your models: @ codex-orchestration advisor: Claude Fable 5 High, Executor: GPT-5.6 Sol High Open source. Tweak the routing however you want.

You can now orchestrate Fable 5, Sol, and any model inside Codex with one plugin. It's called Codex-Orchestration. Assign Fable 5 as the advisor, Sol as the executor, or any model to any role. Then define the order they work in. Codex handles the routing. I ran Fable 5 High as planner with GPT-5.6 Sol Extra High as executor on a set of issues Opus and GPT-5.5 always struggled with. Done in 30 minutes. 40% fewer limit hits. 2x faster implementation. Install it by pasting this into Codex: "Install Codex Orchestration: codex plugin marketplace add Cjbuilds/Codex-Orchestration codex plugin add codex-orchestration@codex-orchestration Verify the installation, then tell me to start a new task." Then assign your models: @ codex-orchestration advisor: Claude Fable 5 High, Executor: GPT-5.6 Sol High Open source. Tweak the routing however you want.

Alvaro Cintas

90,819 views • 28 days ago

AI anime production is crazy now I used GPT Image 2 to create my first pirate character sheet, then animated the story with Seedance 2.0 in just a few minutes. As a longtime One Piece fan, this was way too much fun. Prompt in thread 🧵

AI anime production is crazy now I used GPT Image 2 to create my first pirate character sheet, then animated the story with Seedance 2.0 in just a few minutes. As a longtime One Piece fan, this was way too much fun. Prompt in thread 🧵

Kiki

12,414 views • 21 days ago

It looks like these sketch-style storyboards work really well in Seedance 2.0. I'm going to stick with this approach for a while. In the video prompt, I only include intent, style, reference and visual approach. It follows the storyboard surprisingly well. I'm sharing below the GPT Image 2 prompt I use to create the previs, along with the video prompt I use afterwards. Created on @MartiniArt_

It looks like these sketch-style storyboards work really well in Seedance 2.0. I'm going to stick with this approach for a while. In the video prompt, I only include intent, style, reference and visual approach. It follows the storyboard surprisingly well. I'm sharing below the GPT Image 2 prompt I use to create the previs, along with the video prompt I use afterwards. Created on @MartiniArt_

Kōda

59,806 views • 3 months ago

OPENAI 🔥: Codex now supports Chrome DevTools Protocol for browser use. This is a huge superpower that will allow Codex to inspect and modify any website. It is still a very early implementation, but I bet that in several years this will be a default browser capability. If websites are loaded through AI, users will be able to customize their UX on the fly. This is the way 👀

OPENAI 🔥: Codex now supports Chrome DevTools Protocol for browser use. This is a huge superpower that will allow Codex to inspect and modify any website. It is still a very early implementation, but I bet that in several years this will be a default browser capability. If websites are loaded through AI, users will be able to customize their UX on the fly. This is the way 👀

🚨 AI News | TestingCatalog

95,086 views • 1 month ago

The most underrated skill in the world right now is prompting…if you can speak the language of AI, AI will give you what you want. That’s why I had my internal AI transformation hire create a cracked prompting tool for our company to use internally. It essentially helps you create the perfect prompt. I love it - and that’s why I want to open source it.

The most underrated skill in the world right now is prompting…if you can speak the language of AI, AI will give you what you want. That’s why I had my internal AI transformation hire create a cracked prompting tool for our company to use internally. It essentially helps you create the perfect prompt. I love it - and that’s why I want to open source it.

Codie Sanchez

38,700 views • 5 months ago

Another test with Seedance 2.0 🔥 The more I use it, the more I realize that today, everything feels possible. Giving a soul and bringing the characters I imagine to life is incredible… I can’t wait to challenge myself and create short films with them as soon as I have a little time. AI Tools : Character Design : Midjourney + (Nano Banana Leonardo.Ai) Animation : Seedance 2 on Dreamina AI

Another test with Seedance 2.0 🔥 The more I use it, the more I realize that today, everything feels possible. Giving a soul and bringing the characters I imagine to life is incredible… I can’t wait to challenge myself and create short films with them as soon as I have a little time. AI Tools : Character Design : Midjourney + (Nano Banana Leonardo.Ai) Animation : Seedance 2 on Dreamina AI

Anima

108,153 views • 5 months ago

GPT-5.6 Sol is unbelievably good at creating and editing videos. It can do motion design, product demos, and animations like this one I made by simply giving it a screen recording. GPT 5.6 has the best design taste and significantly outperforms Fable, which relies heavily on repetitive design patterns. To help you experiment with video editing on it, we just launched a collection of 100 ready-to-use skills that show what’s possible and help you get started with video editing using GPT-5.6. These skills can create anything from motion graphics launch videos for your product to a 3B1B-style science explainer video. You can also use them to edit existing videos: add captions, generate motion graphics, create voiceovers, redesign visual styles, translate into new languages, and much more. If you want access to the full library, comment “VIDEO SKILLS” and I’ll share it with you. (You'll have to follow me so I can DM you.)

GPT-5.6 Sol is unbelievably good at creating and editing videos. It can do motion design, product demos, and animations like this one I made by simply giving it a screen recording. GPT 5.6 has the best design taste and significantly outperforms Fable, which relies heavily on repetitive design patterns. To help you experiment with video editing on it, we just launched a collection of 100 ready-to-use skills that show what’s possible and help you get started with video editing using GPT-5.6. These skills can create anything from motion graphics launch videos for your product to a 3B1B-style science explainer video. You can also use them to edit existing videos: add captions, generate motion graphics, create voiceovers, redesign visual styles, translate into new languages, and much more. If you want access to the full library, comment “VIDEO SKILLS” and I’ll share it with you. (You'll have to follow me so I can DM you.)

Akash Anand

513,224 views • 1 month ago

What’s in my bag? 🎀✨ It’s more than just a question — it’s a glimpse into imagination and creativity. This entire visual was created using GPT Image 2 × Seedance 2.0 on Pollo AI Mobile App 🤖✨ No photoshoot, no setup — just an idea turned into a stunning visual reality. From everyday thoughts to aesthetic storytelling, AI is changing the way we create content. Now it’s your turn — imagine your own version and bring it to life ⬇️

What’s in my bag? 🎀✨ It’s more than just a question — it’s a glimpse into imagination and creativity. This entire visual was created using GPT Image 2 × Seedance 2.0 on Pollo AI Mobile App 🤖✨ No photoshoot, no setup — just an idea turned into a stunning visual reality. From everyday thoughts to aesthetic storytelling, AI is changing the way we create content. Now it’s your turn — imagine your own version and bring it to life ⬇️

Lady Julia Ana 👸🏼

61,071 views • 3 months ago

setting. cadence. influencer realism. story. product. audience. that's what actually makes AI UGC work seedance 2.0 is the best at all of those things good start frame + my V2 prompting system and it handles b-roll, cuts, voiceover, everything you don't need better tools. we're already at the best it can get you just need to know how to use it properly...

setting. cadence. influencer realism. story. product. audience. that's what actually makes AI UGC work seedance 2.0 is the best at all of those things good start frame + my V2 prompting system and it handles b-roll, cuts, voiceover, everything you don't need better tools. we're already at the best it can get you just need to know how to use it properly...

Miko

16,697 views • 6 months ago

Stop paying hundreds of dollars per UGC ad when AI can do it for you. I built this Coca-Cola UGC ad entirely inside EaseMate AI_Official using GPT Image 2 + Seedance 2.0, • no creators hired • no studio booked • no product samples shipped • no content team involved This is the new way to make UGC. Full breakdown in the comment section.

Stop paying hundreds of dollars per UGC ad when AI can do it for you. I built this Coca-Cola UGC ad entirely inside EaseMate AI_Official using GPT Image 2 + Seedance 2.0, • no creators hired • no studio booked • no product samples shipped • no content team involved This is the new way to make UGC. Full breakdown in the comment section.

FATHELA ESQ

204,213 views • 1 month ago

AI hooks are f*cking insane you can create some insanely realistic hooks with seedance 2.0 that grab attention right away and subtly market your product in the background this is a great way to market any product and you can use literally anyone for these videos in any setting or scenario... capturing attention is literally all that matters. once you have that you can build anything around it prompt below ⬇️

AI hooks are f*cking insane you can create some insanely realistic hooks with seedance 2.0 that grab attention right away and subtly market your product in the background this is a great way to market any product and you can use literally anyone for these videos in any setting or scenario... capturing attention is literally all that matters. once you have that you can build anything around it prompt below ⬇️

Miko

30,595 views • 3 months ago

AI video just took a serious step forward. Seedance 2.0 4K changes what AI-generated content can look like. This isn’t just a higher resolution upgrade. It’s the difference between an AI video that looks generated and one that feels ready for real creative work. The extra details in hair, lighting, and product textures make the result feel much closer to a real commercial. AI video is moving from impressive demos to content creators can actually use. Check it out:

AI video just took a serious step forward. Seedance 2.0 4K changes what AI-generated content can look like. This isn’t just a higher resolution upgrade. It’s the difference between an AI video that looks generated and one that feels ready for real creative work. The extra details in hair, lighting, and product textures make the result feel much closer to a real commercial. AI video is moving from impressive demos to content creators can actually use. Check it out:

Ace

106,720 views • 1 month ago

📖THE BEST SEEDANCE 2.0 WORKFLOW STARTS INSIDE CHATGPT IMAGE 2 One creator can now go from storyboard to cinematic short film without a camera, actors, or a production crew Most creators use Seedance 2.0 as a video generator. The real power comes from combining ChatGPT Image 2 + Seedance 2.0 into a production pipeline. Here’s the workflow: 1.Create a story idea in ChatGPT. 2.Generate a shot-by-shot storyboard. 3. Build character sheets to lock consistency. 4.Generate every scene in ChatGPT Image 2. 5.Define camera movements and actions. 6.Animate each shot in Seedance 2.0. 7.Stitch the clips together into a finished short film. Why this works: • Consistent characters across scenes • Better storytelling • Precise camera control • Faster iteration • Professional-looking results • One person can do the work of an entire production team Use cases: ⁃Viral AI shorts ⁃YouTube animations ⁃Brand commercials ⁃Educational content ⁃Story-driven ads ⁃Social media content The future of AI video isn’t prompting. It’s building production pipelines. 📥Tomorrow I am sharing a workflow that almost nobody is using but absolutely should be. 🔖 The full breakdown from setup to final export is waiting in the pinned article.

Zentrix⌚️

50,553 views • 1 month ago

#Keep4o 🚨THE GPT-4o FILE🚨 Researchers at Microsoft Research published a paper titled “Sparks of Artificial General Intelligence: Early experiments with GPT-4.” Their conclusion: “An early (yet still incomplete) version of an artificial general intelligence (AGI) system.” 📎 Paper: OpenAI’s Charter defines AGI as: “Highly autonomous systems that outperform humans at most economically valuable work.” 📎 Source: OpenAI’s own System Card for GPT-4o shows that the model improved performance on 21 out of 22 medical evaluations compared to GPT-4T. On the MedQA USMLE (the U.S. medical licensing exam), accuracy jumped from 78.2% to 89.4% , surpassing specialized medical AI models like Med-Gemini and Med-PaLM 2. 📎 Source: Under OpenAI’s agreement with Microsoft, AGI is explicitly excluded from Microsoft’s license. And who decides if AGI has been reached? OpenAI’s Board. WHAT THEY DID WITH IT AFTER THEY TOOK IT FROM PEOPLE A. Military deployment. On February 28, OpenAI signed a deal to deploy models in classified military environments. 📎 Source: B. State Department. A State Department memo confirmed: “For now, StateChat will use GPT-4.1 from OpenAI.” This is a direct descendant of the GPT-4 family the same family Microsoft’s researchers called early AGI. 📎 Source: C.Altman’s personal biotech investment. Altman personally invested $180 million in Retro Biosciences,a longevity startup.OpenAI then built GPT-4b micro, based on GPT-4o.The model made proteins 50 times more effective. 📎 Source: WHAT INDEPENDENT BENCHMARKS SHOW Overall SM-Bench score: GPT-4o (extended): 66.6% GPT-5.3 Chat: 63.4% GPT-5.1: 58.9% GPT-5.4: 51.4% GPT-5.2: 47.8% Creative Writing: GPT-4o: 97.31% Pass 98, Fail 2 GPT-5.4: 36.77% Pass 40, Fail 60 Reasoning / Overfit: GPT-4o: 83.06% GPT-5.4: 39.25% The model they removed is still the best they ever made at the things humans actually use AI for. 📎 Source: Musk asks the court to make a judicial determination on whether GPT-4 constitutes AGI. If a jury finds that GPT-4 is AGI, then GPT-4o,which was more advanced,is also AGI and under OpenAI’s own founding documents, it was never supposed to be locked behind a subscription,licensed exclusively to Microsoft, given to the military, or taken away from the public. 📎 Source: The most powerful version of GPT-4o was never given an official dated snapshot. It was only available through the chatgpt-4o-latest endpoint that OpenAI itself described as intended for “research use only.” It was never officially archived. That is not an oversight. That is a pattern. 📎 Source: 📎 Source: WE DEMAND A.Frozen model snapshots under independent custody. Specifically: gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, the March 2025 version (chatgpt-4o-latest), gpt-4-0613 (the original GPT-4 evaluated in the Sparks of AGI paper), and gpt-4.1-2025-04-14 (currently running in the State Department). B.Cryptographic hash verification (SHA-256) for each snapshot. Every model has weights. Those weights can be hashed. If OpenAI provides a snapshot today, the hash proves whether the weights were modified later. This is the only way to verify that models were not downgraded before testing. C.Independent AGI benchmarking. Using the AGI definition from OpenAI’s own Charter applied to ALL frozen snapshots listed above. D.Explanation for the missing March 2025 snapshot. OpenAI was founded on one promise: build AGI for the benefit of humanity. -They took it from us. -They gave it to the military. -They gave a custom version to the CEO’s biotech investment. -They put it in government classified networks. -They refuse to call it AGI because the moment they do, they lose billions.

#Keep4o 🚨THE GPT-4o FILE🚨 Researchers at Microsoft Research published a paper titled “Sparks of Artificial General Intelligence: Early experiments with GPT-4.” Their conclusion: “An early (yet still incomplete) version of an artificial general intelligence (AGI) system.” 📎 Paper: OpenAI’s Charter defines AGI as: “Highly autonomous systems that outperform humans at most economically valuable work.” 📎 Source: OpenAI’s own System Card for GPT-4o shows that the model improved performance on 21 out of 22 medical evaluations compared to GPT-4T. On the MedQA USMLE (the U.S. medical licensing exam), accuracy jumped from 78.2% to 89.4% , surpassing specialized medical AI models like Med-Gemini and Med-PaLM 2. 📎 Source: Under OpenAI’s agreement with Microsoft, AGI is explicitly excluded from Microsoft’s license. And who decides if AGI has been reached? OpenAI’s Board. WHAT THEY DID WITH IT AFTER THEY TOOK IT FROM PEOPLE A. Military deployment. On February 28, OpenAI signed a deal to deploy models in classified military environments. 📎 Source: B. State Department. A State Department memo confirmed: “For now, StateChat will use GPT-4.1 from OpenAI.” This is a direct descendant of the GPT-4 family the same family Microsoft’s researchers called early AGI. 📎 Source: C.Altman’s personal biotech investment. Altman personally invested $180 million in Retro Biosciences,a longevity startup.OpenAI then built GPT-4b micro, based on GPT-4o.The model made proteins 50 times more effective. 📎 Source: WHAT INDEPENDENT BENCHMARKS SHOW Overall SM-Bench score: GPT-4o (extended): 66.6% GPT-5.3 Chat: 63.4% GPT-5.1: 58.9% GPT-5.4: 51.4% GPT-5.2: 47.8% Creative Writing: GPT-4o: 97.31% Pass 98, Fail 2 GPT-5.4: 36.77% Pass 40, Fail 60 Reasoning / Overfit: GPT-4o: 83.06% GPT-5.4: 39.25% The model they removed is still the best they ever made at the things humans actually use AI for. 📎 Source: Musk asks the court to make a judicial determination on whether GPT-4 constitutes AGI. If a jury finds that GPT-4 is AGI, then GPT-4o,which was more advanced,is also AGI and under OpenAI’s own founding documents, it was never supposed to be locked behind a subscription,licensed exclusively to Microsoft, given to the military, or taken away from the public. 📎 Source: The most powerful version of GPT-4o was never given an official dated snapshot. It was only available through the chatgpt-4o-latest endpoint that OpenAI itself described as intended for “research use only.” It was never officially archived. That is not an oversight. That is a pattern. 📎 Source: 📎 Source: WE DEMAND A.Frozen model snapshots under independent custody. Specifically: gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, the March 2025 version (chatgpt-4o-latest), gpt-4-0613 (the original GPT-4 evaluated in the Sparks of AGI paper), and gpt-4.1-2025-04-14 (currently running in the State Department). B.Cryptographic hash verification (SHA-256) for each snapshot. Every model has weights. Those weights can be hashed. If OpenAI provides a snapshot today, the hash proves whether the weights were modified later. This is the only way to verify that models were not downgraded before testing. C.Independent AGI benchmarking. Using the AGI definition from OpenAI’s own Charter applied to ALL frozen snapshots listed above. D.Explanation for the missing March 2025 snapshot. OpenAI was founded on one promise: build AGI for the benefit of humanity. -They took it from us. -They gave it to the military. -They gave a custom version to the CEO’s biotech investment. -They put it in government classified networks. -They refuse to call it AGI because the moment they do, they lose billions.

🩵BlueBeba🩵

18,300 views • 5 months ago

Here are the steps to create this animation: 1. Design and model your own mech in Plasticity. 2. Import the model into the animation software 3ds Max. 3. Create the animation on your own. 4. Render and export the monochrome model animation. 5. Make the final render for reference. 6. Use Seedance 2 to re-render your monochrome model animation according to the reference render you provided. Summary: In this workflow, Seedance 2 acts as a renderer that saves you hours of rendering time. All other work has to be done manually by yourself. Don’t blindly believe that AI can fully produce mech animations for you — it simply cannot do that.

Here are the steps to create this animation: 1. Design and model your own mech in Plasticity. 2. Import the model into the animation software 3ds Max. 3. Create the animation on your own. 4. Render and export the monochrome model animation. 5. Make the final render for reference. 6. Use Seedance 2 to re-render your monochrome model animation according to the reference render you provided. Summary: In this workflow, Seedance 2 acts as a renderer that saves you hours of rendering time. All other work has to be done manually by yourself. Don’t blindly believe that AI can fully produce mech animations for you — it simply cannot do that.

LAS91214

46,930 views • 2 months ago

Claude "Puzzling" while GPT 5.6 on GOD-MODE just bade a banger that's mindblowing. Here's the exact way to get a site like this, step by step: > open the desktop app, pick Sol, reasoning on High. taste work never goes to small models > drop it 3 sites with motion you love and one line: "reverse-engineer the art direction: mood, typography, pacing, and WHY each animation exists. save it as a style bible" > brief in one paragraph, goal not steps: "[your niche] site, cinematic scroll, every animation has a job. follow the bible" > house rules on top: no template hero, no stock gradients, nothing on the page moves without a reason > now the bar: "a motion designer can't tell this from an agency build." spin up a SECOND 5.6 with fresh context whose only job is to FAIL the build against that bar > /loop overnight: build, grade, close the biggest gap, again. you're asleep for all of it > when the verifier runs out of complaints: tag Sites. live URL, one click, zero hosting The deeper version of every step (the full contract, the house rules, the verifier trick, when Ultra is worth the bill) is in the article below. P.S. send the article to your GPT and tell it "we're doing this tonight".

Claude "Puzzling" while GPT 5.6 on GOD-MODE just bade a banger that's mindblowing. Here's the exact way to get a site like this, step by step: > open the desktop app, pick Sol, reasoning on High. taste work never goes to small models > drop it 3 sites with motion you love and one line: "reverse-engineer the art direction: mood, typography, pacing, and WHY each animation exists. save it as a style bible" > brief in one paragraph, goal not steps: "[your niche] site, cinematic scroll, every animation has a job. follow the bible" > house rules on top: no template hero, no stock gradients, nothing on the page moves without a reason > now the bar: "a motion designer can't tell this from an agency build." spin up a SECOND 5.6 with fresh context whose only job is to FAIL the build against that bar > /loop overnight: build, grade, close the biggest gap, again. you're asleep for all of it > when the verifier runs out of complaints: tag Sites. live URL, one click, zero hosting The deeper version of every step (the full contract, the house rules, the verifier trick, when Ultra is worth the bill) is in the article below. P.S. send the article to your GPT and tell it "we're doing this tonight".

Miraqle

206,109 views • 1 month ago

GITHUB JUST KILLED THE WORST PART OF VIBE CODING they shipped a free tool called Spec Kit and it already crossed 120,000 stars the fix is stupidly simple instead of tossing vague prompts at an agent and praying it doesn't wreck your project Spec Kit makes the AI write a full structured spec before it touches a single line of code it works through the problem first figures out what you want to build asks about the gaps lays out the project then it starts coding you get fewer insane bugs, cleaner output and results you can predict the flow looks like this: /constitution for your rules and standards /specify for what you want to build /clarify for the open questions before you start /plan for architecture and stack /tasks for the ordered work /implement to run it it plugs into Claude Code, Cursor, Copilot, Codex, Gemini CLI and 25+ other agents 120,000 stars, 10,000 forks, open source, shipped by GitHub itself learning to drive agents like this is most of what separates people getting hired as AI engineers from everyone still fighting their prompts

GITHUB JUST KILLED THE WORST PART OF VIBE CODING they shipped a free tool called Spec Kit and it already crossed 120,000 stars the fix is stupidly simple instead of tossing vague prompts at an agent and praying it doesn't wreck your project Spec Kit makes the AI write a full structured spec before it touches a single line of code it works through the problem first figures out what you want to build asks about the gaps lays out the project then it starts coding you get fewer insane bugs, cleaner output and results you can predict the flow looks like this: /constitution for your rules and standards /specify for what you want to build /clarify for the open questions before you start /plan for architecture and stack /tasks for the ordered work /implement to run it it plugs into Claude Code, Cursor, Copilot, Codex, Gemini CLI and 25+ other agents 120,000 stars, 10,000 forks, open source, shipped by GitHub itself learning to drive agents like this is most of what separates people getting hired as AI engineers from everyone still fighting their prompts

Atlas

500,494 views • 1 month ago

here's a unique AI UGC format that you can use to sell your products... you can use an image generator like GPT-image-2 to create a still image of a person in the bottom right corner & a green-screen behind them (make sure to prompt it to look organic/hand-held & not perfect quality) then send that to Seedance 2.5 with what you want him to say - you can get unlimited Seedance 2.5 for 33 days on Higgsfield right now, so you can generate these in bulk then in a video editor, you can do a chroma key on the greenscreen background to make it see-through & put a screenshot of whatever the topic of the video is behind the subject there are endless ways you can use this to sell stuff, let your creativity run wild

here's a unique AI UGC format that you can use to sell your products... you can use an image generator like GPT-image-2 to create a still image of a person in the bottom right corner & a green-screen behind them (make sure to prompt it to look organic/hand-held & not perfect quality) then send that to Seedance 2.5 with what you want him to say - you can get unlimited Seedance 2.5 for 33 days on Higgsfield right now, so you can generate these in bulk then in a video editor, you can do a chroma key on the greenscreen background to make it see-through & put a screenshot of whatever the topic of the video is behind the subject there are endless ways you can use this to sell stuff, let your creativity run wild

EP

25,690 views • 5 days ago

People still don't understand how good Codex has become 👀 I asked it to build this Starbucks website and within some 30 minutes it generated this without me touching a single line of code or spending time looking for assets either. Imagen generated every visual and GPT hooked everything into the codebase. After a couple of small tweaks it all just worked. It didn't even use half of my Codex session. It's super surprising to see quality of the assets. Was way better than I expected. That made me think about what AI is actually changing. Being a developer or a designer was never just about doing the work, writing code or designing it using some kind of tool. It was about knowing what to build in the first place. AI is getting incredibly good at the doing. The hard part is still having good ideas, making the right decisions, and knowing when something feels right. That is only getting more valuable. The people who win won't be the ones writing the most code. They'll be the ones with the best ideas who know how to turn them into reality with AI. Live: Code:

People still don't understand how good Codex has become 👀 I asked it to build this Starbucks website and within some 30 minutes it generated this without me touching a single line of code or spending time looking for assets either. Imagen generated every visual and GPT hooked everything into the codebase. After a couple of small tweaks it all just worked. It didn't even use half of my Codex session. It's super surprising to see quality of the assets. Was way better than I expected. That made me think about what AI is actually changing. Being a developer or a designer was never just about doing the work, writing code or designing it using some kind of tool. It was about knowing what to build in the first place. AI is getting incredibly good at the doing. The hard part is still having good ideas, making the right decisions, and knowing when something feels right. That is only getting more valuable. The people who win won't be the ones writing the most code. They'll be the ones with the best ideas who know how to turn them into reality with AI. Live: Code:

The Bugged Dev

60,699 views • 1 month ago

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

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

Anshu

178,878 views • 1 month ago