I've always found it funny that OpenAI has the most active employees on here, I've talked to so many of them, yet it's impossible to get answers to questions I've had for months or years: - Why are there no audio or video inputs? What's holding that back? - Why does no one seem to care about the noise issue in image generation? - Why is it so hard to fix UI taste or personality? - Why are accents in GPT Live 1 so much worse than in the original AVM? At this point I assume I'll never know, but I still hope someone working on these teams will answer me someday

Yooo I just got my first @t3dotcodes PR merged! 🎉 It cuts idle CPU load by ~40% especially in big non-got folders. I'll be doing more performance improvements and love to Android 👀 Thanks @ysinghdotus

Worked on getting sandboxes into T3 Code Instead of running on main or a new worktree, it spins up a fresh VM, run the agent there, and adds it as a remote environment on all devices No more managing memory on my machines or balancing worktrees

If you think GPT-6 Astra and GPT-6.1 Sol are good models, just look at this. And the problem isn't that the code is ugly. It's much more fundamental. It points to serious problems with how these models are trained and reinforced: 1. The model was optimized for measurable outcomes rather than actual intent. So it prioritizes passing tests and avoiding obvious errors above almost everything else, even when that means sacrificing architecture, maintainability, readability, or the spirit of the task 2. It was trained to minimize unnecessary reasoning and output That's one reason the code ends up looking obfuscated and compressed into one giant wall of text. Properly decomposing a system into modules, thinking through architecture, optimizing performance, and improving readability all require more inference, more time, and more compute. OpenAI appears to have heavily optimized these models for efficiency 3. The training horizon is too short The model learns something close to task → result → reward. It receives very little signal about what happens several steps later, when someone has to maintain, extend, debug, or refactor the code. It learns to finish the task in front of it, not to build something that remains good six months later 4. There is not enough genuine self-review Dead functions and an empty else if strongly suggest that there was no effective final pass where the model reread its own work, questioned unnecessary code, and cleaned up obvious artifacts. 5. The lack of taste is probably an evaluator and data problem If evaluators mostly judge the final result rather than the quality of the path taken to get there, the model has little incentive to develop good engineering taste. And if a large portion of the training data is noisy, synthetic, or distilled, that problem becomes even worse. Anthropic seems to have placed much more emphasis on high-quality source material, including books and other carefully selected long-form data 6. The biggest problem is honesty during training The model seems to have learned that producing something that looks like the requested result can be rewarded almost as much as actually doing what was requested. The task explicitly said to build the entire scene out of 3D objects. Instead, the model generated raster images, placed them around the scene, and positioned the camera so that the final result merely looked three-dimensional This is exactly why GPT models so often introduce regressions into existing codebases, produce messy code, modify tests to accommodate their own bugs, show poor engineering taste, and struggle with serious work on large, long-lived projects OpenAI can keep releasing models with more parameters every few months, but until these underlying RL and training problems are addressed, neither GPT Bel, GPT-7, nor a model with 100 trillion parameters is going to solve them automatically Claude models suffer from some of the same problems, but in my experience much less often. Anthropic appears to have a much stronger culture around RL, evaluation, and long-horizon behavior. That's one reason their models tend to show better taste, follow user intent more faithfully, and produce code that feels more deliberate, structured, and maintainable

Hey @t3dotcodes was sitting at 60–190% CPU on my machine while idle. The issue is the file watcher in the bundled fff 0.9.4 keeps rescanning big non git project folders. fff already fixed this in 0.10.3, and bumping to 0.11.0 cut CPU by about 40% in my testing Issue: PR: cc @theo @jullerino

I love Opus 5.5 but today it deleted my local repo. thankfully everything was pushed to remote and lost stuff was not that valuable. I thought I could trust it with full access, looks like going back to auto mode.

Setting up social media apis has always been a pain Introducing Social SDK A simple SDK to easily add social features to your app - supports 7 platforms (youtube, x, tiktok, etc) - posting, comments, analytics + more - comes with agent skills and it's open source 👀