Microsoft MAI-Thinking-1
MAI-Thinking-1
Microsoft's first in-house reasoning model, available only through its own cloud.
Updated
The read
MAI-Thinking-1 is Microsoft building its own frontier model rather than relying only on OpenAI's. It is a sparse model of about a trillion parameters with 35 billion active, text only, with a 256,000-token window, and it has been in public preview in Microsoft Foundry since August. Microsoft's own figures on coding and competition maths are strong, but neither major independent leaderboard has rated it yet, so the score here is provisional and deliberately cautious.
Where it shines
- Strong published scores on coding and competition maths
- Runs inside Microsoft Foundry with Azure's compliance controls
- Function calling supported from launch
Worth knowing first
- No independent leaderboard has rated it yet
- Text only, and available only through Microsoft's cloud
Index score
76.0
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 61 of 73
- Builds
- LLMs
- Category
- Reasoning
- Output
- Text in, text and tool calls out
- Pricing
- Paid only
- Best for
- Azure teams who want a reasoning model without a new vendor
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
Also worth a look
Full index- 21GPT-6 AstraGPT-6 AstraProvisionalOpenAI's most capable model: a specialist for the hardest problems, priced like one.Reasoning modelPaid only84.9
- 25GPT-5.6 SolGPT-5.6 SolProvisionalStrong and now discounted, but GPT-6.1 Sol beats it for half the money.Reasoning modelPaid only84.4
- 28DeepSeek V4 ProDeepSeek V4 ProProvisionalDeepSeek's reasoning flagship, now out of preview and outscored by its own Flash model.Reasoning modelFree tier83.4
