MiMo-V2.5-Pro

MiMo-V2.5-Pro is a reasoning model from Xiaomi in the MiMo-V2.5 family. 27 benchmarks count toward its score, in 6 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.58.1 ±4.6
CoverageShare of the index weight with results.85%
SpeedOutput tokens per second.29/s
Input / 1MUS dollars per 1M input tokens.$0.435
Output / 1MUS dollars per 1M output tokens.$0.87
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1465 (#32)

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

60,919 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
58.6
CodingCode writing and repair.
59.1
ReasoningLogic problems and puzzles.
55.1
MultimodalTasks with images and text.
N/A
KnowledgeFacts and expert knowledge.
55.6
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
70.2
MathMath problems.
58.3

Results

27 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.
τ²-Bench Tool-Agent-User EvaluationAgentic94.2%66.4Victor Barres et al.
Artificial Analysis GPQA DiamondKnowledge86.6%57.3Artificial Analysis
MMLU ProKnowledge84.6%53.81 Sept 2026Vals AI
GPQA DiamondKnowledge82.6%54.51 Sept 2026Vals AI
LiveCodeBenchCoding81.4%58.31 Sept 2026Vals AI
Artificial Analysis IFBenchInstruction79.9%70.2Artificial Analysis
Artificial Analysis Long Context ReasoningReasoning79.7%63.4Artificial Analysis
SWE-benchCoding74.0%56.91 Sept 2026Vals AI
τ³-Bench Tool-Agent-User EvaluationAgentic72.9%57.1Sierra Research
Claw-EvalAgentic63.8%58.7Bowen Ye et al.
Gert Labs Composite Game BenchmarkAgentic62.7%68.1Gert Labs
Artificial Analysis Coding IndexCoding60.2%61.4Artificial Analysis
Terminal-Bench 2.1Agentic57.3%57.921 Sept 2026Vals AI
SWE-bench ProCoding57.2%59.3Xiang Deng et al.
Artificial Analysis SciCodeCoding50.6%63.1Artificial Analysis
Humanity's Last ExamKnowledge48.0%69.4Center for AI Safety et al.
Artificial Analysis Humanity's Last ExamKnowledge35.7%63.3Artificial Analysis
Vibe Code Bench v1.1Coding34.1%56.4OpenHands21 Sept 2026Vals AI
Humanity's Last Exam without toolsKnowledge34.0%57.6OpenAI
GDPval-AA normalizedAgentic30.4%58.9Artificial Analysis
Artificial Analysis Intelligence IndexKnowledge26.0%54.9Artificial Analysis
Artificial Analysis Agentic IndexAgentic22.7%55.6Artificial Analysis
Artificial Analysis Omniscience AccuracyKnowledge22.4%41.6Artificial Analysis
ProofBench v1.1Math22.0%58.321 Sept 2026Vals AI
Code MigrationCoding21.6%58.721 Sept 2026Vals AI
Critical Physics TasksReasoning4.0%46.8Artificial Analysis
APEX-Agents-AAAgentic2.4%43.0Artificial Analysis / Mercor
Agent Arena steerabilityAgentic-4.062.915 Sept 2026LMArena
Agent Arena command recoveryAgentic-4.961.815 Sept 2026LMArena
Agent Arena task outcomeAgentic-9.057.215 Sept 2026LMArena

27 benchmarks count, from 30 of 30 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 authorsOpenRouter, collected directlyNo licence statedVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AILMArena, collected directlyCC BY 4.0 (lmarena-ai/leaderboard-dataset on Hugging Face)

Same level, lower price

Qwen3.8-Flash-Next66.4 · Freedots3-note Preview65.5 · FreeApodex 1.1 Mini59.1 · Free

More from Xiaomi

MiMo-V2.6-Pro73.5MiMo-V2.555.7MiMo-V2-Pro53.6MiMo-V2-Omni49.1MiMo-V2-Flash43.9