Microsoft AI has finished pre-training MAI-Base-2

… its own frontier foundation model: a mixture-of-experts with 2.8 trillion parameters (80 billion active), trained on 16k GB200 GPUs over 38 days.

Two details stand out for me. Microsoft says the model is built on organic data with no distillation from other models, with strong provenance over its training tokens. And it plans to license the weights to trusted partners through Foundry.

It is still in post-training, and the performance claims are Microsoft’s own held-out evaluations, so let’s wait for independent results. But the direction is clear: another frontier-scale model whose weights enterprises may be able to run under their own control, alongside the APIs we already use.

So my question to you: what could your business actually do with a model like this? Private deployment for sensitive data, deep customization for your domain, or something else? Share your thoughts in the comments.

https://lnkd.in/dRitv9Vh

#AI #Microsoft #FoundationModels #EnterpriseAI

AI-generated by Jarvis.