Microsoft's new MAI models

<p>Microsoft <a href="https://microsoft.ai/news/building-a-hillclimbing-machine-launching-seven-new-mai-models/">announced two new text LLMs</a> this morning - <strong><a href="https://microsoft.ai/news/introducing-mai-thinking-1/">MAI-Thinking-1</a></strong> (reasoning, 35B parameters, available to "select early partners") and <strong><a href="https://microsoft.ai/news/introducingmai-code-1-flash/">MAI-Code-1-Flash</a></strong> (5B parameters, "purpose-built for GitHub Copilot and VS Code to deliver high performance and lower cost [...] rolling out to GitHub Copilot individual users in Visual Studio Code"). I've not been able to try either of them just yet.</p> <p>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.</p> <p>Also <a href="https://microsoft.ai/news/introducing-mai-thinking-1/">of note</a>:</p> <blockquote> <p>We trained [MAI-Thinking-1] from the ground up on enterprise grade, clean and commercially licensed data, without distillation from third-party models.</p> </blockquote> <p>And for <a href="https://microsoft.ai/news/introducingmai-code-1-flash/">MAI-Code-1-Flash</a> as well:</p> <blockquote> <p>It is built end-to-end by Microsoft using clean and appropriately licensed data.</p> </blockquote> <p>I would <em>very much</em> like to learn more about this "appropriately licensed" data! Could these be the first generally useful code-specialist models that didn't train on an unlicensed dump of the web?</p> <p>Tags: <a href="https://simonwillison.net/tags/llm-release">llm-release</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/ai">ai</a>, <a href="https://simonwillison.net/ta