Microsoft's new MAI models
Read OriginalThis article covers Microsoft's announcement of two new text LLMs: MAI-Thinking-1 (1T parameters, 35B active) and MAI-Code-1-Flash (137B parameters, 5B active) for GitHub Copilot and VS Code. It discusses their low parameter counts, performance claims against Sonnet 4.6, and training on clean data. The author corrects initial errors about model sizes and notes that despite claims, the training data still relies on web crawls with licensing issues, similar to other major LLMs. Includes updates on technical paper details.
Comments
No comments yet
Be the first to share your thoughts!
Browser Extension
Get instant access to AllDevBlogs from your browser
Top of the Week
No top articles yet