I wasn't able to train a pixel art model from scratch (at least not without spending over $2k), so I'm making the effort somewhat worth it by releasing my synthetic dataset instead, over 1.4M images generated with Krea 2 and a custom LoRA I also have some ideas for improving asset quality a lot The main issue right now is removing the magenta background, but I think I know how to fix it You can check it out here

Meta launched two things yesterday : a state-of-the-art model & a new pricing system for foundation models. Muse Spark propels US open source models to the frontier. Meanwhile, the pricing system resets the industry’s economics.

Claude Fable 5 is now available in Computer as an orchestrator model. This is Anthropic's state-of-the-art model for long, complex tasks. Available only to Pro and Max subscribers in Computer.

After much testing and talking to actual users across all three proactive AI agents: Muse - best one for your parents, heavy IG/FB users, people who like or need easy setups, people who don’t want to build anything, best use case idea generator by a mile (aka great if you don’t know what you want to use it for but you have to click the idea tab!), best for people who still say “wtf can AI do for me” Dots - best for OpenAI users, people who might need to build/code, people who deal with slack a lot, people who love voice features Instinct - people who live in their iMessage/don’t want to open another app, people who don’t need crazy fancy visualizations, people who want internet watchdogs (like “next time AJ Jacobs writes an article, send it to me”), easiest setup (tied with Muse) I am a very heavy Instinct user and light Muse/Dots user, largely because it's in iMessage. Muse also made several basic mistakes when I was chatting with it and forgot things after awhile. Didn’t feel like a state of the art model, despite somewhat equal benchmarks to Astra on GDPval-AA, but that might change with today's new model news, so I'm excited to test. And my honest take: still feels just very early across the space. We're going to see more. Note: answers subject to change, literally day by day.