Research2026-10-15

Microsoft AI CEO Mustafa Suleyman said its new MAI-Base-2 model beats OpenAI’s GPT 5.6 Sol. It used one quarter of the compute OpenAI used. It also used twelve times less compute than DeepSeek’s V4-Pro-Base. Microsoft tested it privately against OpenAI models. The model was announced on the same day. Microsoft has a major partnership with OpenAI. The relationship between the two has frayed recently. Microsoft is now building its own models more aggressively.

What it unlocks

Training a stronger base model with far less computing power.

  • 1/4 the compute of OpenAI’s GPT 5.6 Sol model
  • 12x less compute than DeepSeek-V4-Pro-Base

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.