Microsoft model product
MAI-Thinking-1 is Microsoft AI's first in-house reasoning model, a 35B-active / ~1T-total parameter sparse Mixture of Experts model (base model MAI-Base-1) trained from scratch without distillation from third-party models. Built with Microsoft's Hill-Climbing Machine pipeline, it was pre-trained on 30T tokens of clean, commercially licensed, human-generated data (plus 3.55T mid-training tokens), then post-trained via reinforcement learning across STEM, agentic coding, and helpfulness/safety specialists consolidated into a single model. It delivers strong mathematical reasoning and software-engineering performance for its weight class, going toe-to-toe with Claude Opus 4.6 on SWE-Bench Pro and reaching 97.0% on AIME 2025. It supports a 256k token context window, function calling, and developer instructions, and is preferred over Claude Sonnet 4.6 in blind human side-by-side evaluations.
Updated Aug 10, 2026. Default version: MAI-Thinking-1
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version MAI-Thinking-1.
| AdvancedIF | 0.8 | 1 | 2 | 100.0% | C | |
| AIR-Bench | 0.9 | 1 | 1 | 100.0% | C | |
| CorpusQA | 0.8 | 1 | 1 | 100.0% | C | |
| CyberSecEval 4 | 0.6 | 1 | 1 | 100.0% | C | |
| GraphWalks | 0.9 | 1 | 3 | 100.0% | C | |
| LongFact | 1.0 | 1 | 1 | 100.0% | C | |
| SimpleQA Verified | 0.3 | 1 | 1 | 100.0% | C | |
| TruthfulQA | 0.9 | 1 | 18 | 100.0% | C | |
| BFCL-v3 | 0.7 | 4 | 19 | 83.3% | C | |
| AIME 2026 | 0.9 | 5 | 17 | 75.0% | C | |
| LiveCodeBench v6 | 0.9 | 6 | 53 | 90.4% | C | |
| LongBench v2 | 0.6 | 7 | 17 | 62.5% | C | |
| HMMT Feb 26 | 0.8 | 8 | 11 | 30.0% | C | |
| HealthBench Professional | 0.3 | 9 | 9 | 0.0% | C | |
| MedXpertQA | 0.4 | 9 | 12 | 27.3% | C | |
| AIME 2025 | 1.0 | 15 | 114 | 87.6% | C | |
| Multi-Challenge | 0.5 | 16 | 29 | 46.4% | C | |
| IFBench | 0.7 | 19 | 28 | 33.3% | C | |
| MMLU-Pro | 0.8 | 21 | 129 | 84.4% | C | |
| SWE-Bench Pro | 0.5 | 37 | 44 | 16.3% | C | |
| SWE-Bench Verified | 0.7 | 41 | 104 | 61.2% | C | |
| Terminal-Bench 2.0 | 0.5 | 42 | 49 | 14.6% | C | |
| GPQA | 0.8 | 60 | 233 | 74.6% | C |
Preference and agent-evaluation signals from published Arena datasets.
The default version has no published Arena rows, or its source alias has not been resolved.
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
The default version has no provider offering with current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| MAI-Thinking-1 | 63.4 | 1T | N/A | N/A | No | Proprietary |
A concise description based on the published model registry.
MAI-Thinking-1 is Microsoft AI's first in-house reasoning model, a 35B-active / ~1T-total parameter sparse Mixture of Experts model (base model MAI-Base-1) trained from scratch without distillation from third-party models. Built with Microsoft's Hill-Climbing Machine pipeline, it was pre-trained on 30T tokens of clean, commercially licensed, human-generated data (plus 3.55T mid-training tokens), then post-trained via reinforcement learning across STEM, agentic coding, and helpfulness/safety specialists consolidated into a single model. It delivers strong mathematical reasoning and software-engineering performance for its weight class, going toe-to-toe with Claude Opus 4.6 on SWE-Bench Pro and reaching 97.0% on AIME 2025. It supports a 256k token context window, function calling, and developer instructions, and is preferred over Claude Sonnet 4.6 in blind human side-by-side evaluations.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about MAI Thinking 1.
MAI Thinking 1's default version was released on Jun 2, 2026.
No official standard PAYG price is currently available for MAI Thinking 1.
MAI Thinking 1 is published under Microsoft in the model registry.
The current registry does not publish a context window for the default version.
No. The default version is not marked as having publicly available weights.
No published provider offering is currently linked to the default version.
Nearby ranked alternatives include Claude Sonnet 3.7, Gemini 2.5 Flash, Nova Pro.