Microsoft's new MAI models (opens in new tab)

Covers 3 stories including Building a hill-climbing machine: Launching seven new MAI modelsCovered by 3 sources including Simon Willison's Newsletter, The Decoder

Microsoft this morning - It's very interesting to see Microsoft releasing models with such low parameter counts, especially given how expensive larger models are to access right now. They claim MAI-Thinking-1 "is preferred to Sonnet 4.6 in our blind human side-by-side evaluations", which is impressive for a 35B model seeing as I frequently run models larger than that on my own laptop. Tags: <a href=" <a href=" <a href=" <a href=" <a href="

Read the original article

Sign in to keep reading the full article.

Sign Up Log In

Covered in 3 articles

Simon Willison's Newsletter·

Running Python code in a sandbox with MicroPython and WASM

Discussed on Substack

The Decoder

·

Microsoft trained its MAI models on unlicensed web data despite promising "enterprise grade, clean and commercially licensed data"

indiehacker.news·

Covered in 3 articles

Running Python code in a sandbox with MicroPython and WASM

Microsoft trained its MAI models on unlicensed web data despite promising "enterprise grade, clean and commercially licensed data"

#063 - Microsoft cut Claude Code internally then shipped a rival, GitHub gives agents their own app