Meta just shipped two things at once: Muse Spark 1. 2, the model, and Muse Code, an agentic harness that wraps around it. The interesting part is that they pull in opposite directions.
Karpathy-sensei gist hit me directly, and i built end to end hosted version of it, after using ad-hoc version myself for a few months. Whole point is to have agent iterate on same files, and reuse past results. It works quite well.
1 points, 0 comments on Hacker News
1 points, 0 comments on Hacker News
1 points, 1 comments on Hacker News
Originally published on the OxygenLabs blog . Every few weeks a new AI model is announced, and the coverage splits into two camps: one says the model is "open", the other says it is "closed". If you are learning to build with AI, that single distinction affects more of your day-to-day work than...