MAI-Thinking-1
- Made by
- Microsoft
- Released
- 12 August 2026
- Licence
- proprietary
- Size
- 1.0T
- Context
- 256K
- Input, per 1M tokens
- $2
- Output, per 1M tokens
- $8
How this score is made
- HealthBench Professional0.35035.0 out of 100runs −2.5 hardadjusted 37.5weight 0.446, supportingmodel cardReported by the maker.length-adjusted (HealthBench Professional length penalty), standard GPT-5.4 grader and OpenAI rubrics, Microsoft AI run; printed at integer precision.
- HealthBenchNo result here, and none is filled in.
- Health Optimization Bench: incretin33.033.0 out of 100runs +4.0 easyadjusted 29.0weight 0.226, supportingofficial leaderboardNear its ceiling (leader 84.9), so it counts for less.
- SAFE-TeenNo result here, and none is filled in.
- Adjusted average of 2 benchmarks, by weight
- 34.6
- Field average, counts for 42.9% with 2 of 4 benchmarks
- 55.3
- Index
- = 43.5
Benchmark by benchmark
HealthBench Professional42 of 37
- 0.703GPT-6 Astra (Anthropic run)
- 0.692Claude Sonnet 5.5
- 0.660Claude Fable 5
- 0.656Claude Opus 5.5
- 0.655GPT-6 Luna (Anthropic run)
- 0.648Claude Haiku 5.5
- 0.642GPT-6.1 Sol
- 0.627GPT-6.1 Terra
- 0.623Muse Wonder Full
- 0.350MAI-Thinking-1
37 modelsFull benchmark
Health Optimization Bench: incretin evidence synthesis15 of 17
- 84.9Claude Fable 5.1
- 83.8Claude Fable 5
- 82.2GPT-6 Astra (max)
- 81.3Grok 4.6
- 78.3Claude Opus 5
- 77.9GPT-5.6 Sol (max)
- 77.8Kimi K3
- 68.4Muse Spark
- 67.8Gemini 3.8 Flash
- 33.0MAI Thinking
17 modelsFull benchmark
Health Optimization Bench14 of 15
- 70.9Claude Fable 5
- 69.3Claude Opus 5
- 66.8Grok 4.6
- 66.6GPT-5.6 Sol (max)
- 59.9Kimi K3
- 57.2Muse Spark
- 47.3Claude Fable 5.1
- 39.7Gemini 3.6
- 35.6Inkling
- 17.5MAI Thinking
15 modelsFull benchmark