MAI-Thinking-1

MAI-Thinking-1 is a reasoning model from Microsoft in the MAI-Thinking family. 9 benchmarks count toward its score, in 4 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.55.1 ±10.4
CoverageShare of the index weight with results.50%
SpeedOutput tokens per second.
Input / 1MUS dollars per 1M input tokens.N/A
Output / 1MUS dollars per 1M output tokens.N/A
ContextMaximum tokens in one request.256K
EloLMArena rating and rank.N/A

The index is a score out of 100. The ± range shows how much it can change.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
N/A
CodingCode writing and repair.
56.1
ReasoningLogic problems and puzzles.
N/A
MultimodalTasks with images and text.
N/A
KnowledgeFacts and expert knowledge.
55.2
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
64.9
MathMath problems.
53.4

Results

9 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result as a score out of 100.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
American Invitational Mathematics Examination 2025Math97.0%52.9Mathematical Association of America
AIME 2026Math94.5%53.8Qwen
Graphwalks BFS 0K-128KReasoning90.0%OpenAI
LiveCodeBench v6Coding87.7%56.8LiveCodeBench maintainers
Massive Multitask Language Understanding ProfessionalKnowledge85.0%54.5Yubo Wang et al.
Instruction Following BenchmarkInstruction85.0%64.9Benchmark authors
Harvard-MIT Mathematics Tournament February 2026Math84.9%53.7Qwen
Graduate-Level Google-Proof Q&AKnowledge84.2%56.0David Rein et al.
GPQA DiamondKnowledge84.2%56.0David Rein et al.
Software Engineering Benchmark VerifiedCoding73.5%56.5Carlos E. Jimenez et al.
SWE-bench ProCoding52.8%55.1Xiang Deng et al.
Measuring Short-Form Factuality in Large Language ModelsKnowledge31.0%Jason Wei et al.

9 benchmarks count, from 10 of 12 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authors

More from Microsoft

Phi-423.3Fara1.5-27BUnrankedFara1.5-4BUnrankedMAI-Code-1.1-FlashUnrankedMAI-Cyber-1-FlashUnranked