MiMo V2.5 Pro — Benchmarks

Benchmark scores for MiMo V2.5 Pro aggregated from public leaderboards (8 boards), with how it ranks among open models and how it compares with its nearest neighbours. See hardware requirements for what you need to run it.

llmrun Composite Score →

116.4

90% range 111.3–121.6 · from 6 boards

Coding115.690% range 106–126
AgentsNot enough data yet
MathNot enough data yet
Science114.690% range 97–133
Reasoning110.890% range 102–120

MiMo V2.5 Pro reasoning benchmarks

Multi-step logic and problem-solving — can the model think through a hard problem rather than recall an answer.

CritPt
CritPt (Critical Points) tests graduate-level physics problem-solving on research-style questions spanning mechanics, electromagnetism, quantum mechanics, and more. It was created by physicists to probe whether a model can reason through multi-step physics rather than pattern-match textbook answers.
4.0%
4.0%
median 1.1%open leader 26.6%
Overall best: GPT 5.6 Sol Max 32.3%
74th percentile · #18 of 66 open · #132 of 219 all
#18 of 66
#132 of 219
Lab-reported
Epoch AI
DTBench
DTBench is a set of handcrafted, expert-validated multiple-choice questions on decision theory — Newcomb-like problems and related puzzles — built by the Conceptual Reasoning Index project with Anthropic. It tests whether a model can follow an abstract argument rather than recall a fact.
84.5%
84.5%
median 67.7%open leader 93.9%
Overall best: Claude Opus 5.5 Max 98.9%
78th percentile · #16 of 68 open · #93 of 214 all
#16 of 68
#93 of 214
Lab-reported
Epoch AI
LMCA
LMCA (Language Model Conceptual Argumentation) scores models on expert-rated arguments across philosophy, decision theory, and AI risk — domains where there is no empirical feedback loop to check an answer against. It is part of the Conceptual Reasoning Index, built with Anthropic by researchers Emery Cooper and Caspar Oesterheld.
29.5%
29.5%
median 24.7%open leader 55.5%
Overall best: Claude Opus 5.5 Max 68.2%
62nd percentile · #21 of 54 open · #118 of 176 all
#21 of 54
#118 of 176
Lab-reported
Epoch AI

MiMo V2.5 Pro coding benchmarks

Writing, editing, and fixing real code — measured by whether the resulting program actually runs and passes tests.

SciCode
SciCode asks a model to write real scientific-computing code — implementing multi-step numerical methods drawn from physics, chemistry, biology, and math research — and checks the output against reference solutions. It was built by a team of scientists and ML researchers to test coding ability on genuine research problems, not typical software tasks.
50.2%
50.2%
median 38.7%open leader 60.9%
Overall best: Claude Opus 5.5 Max 66.9%
86th percentile · #10 of 66 open · #105 of 207 all
#10 of 66
#105 of 207
Lab-reported
Epoch AI
ALE-Bench
ALE-Bench, from Sakana AI and AtCoder, tests long-horizon algorithmic optimisation using problems drawn from AtCoder Heuristic Contests, where there is no single correct answer and solutions are scored on quality. Results are reported as an AtCoder-style performance rating, where higher is better.
899.8
899.8
median 694.1open leader 1524.5
Overall best: GPT 6 Astra Max 2951.3
69th percentile · #14 of 43 open · #63 of 122 all
#14 of 43
#63 of 122
Lab-reported
Epoch AI

Other results (1)· leaderboard-only, not part of the llmrun Score

LMArena WebDev
Code Arena (formerly WebDev Arena) is Arena's head-to-head coding leaderboard: two models each build a real web app or UI from the same prompt, and people vote blind on which result is better. The rating shown is a relative Bradley-Terry score, not a percentage — compare it only to other models on this board.
1481.3
1481.3
median 1411.4open leader 1655.2
Overall best: Claude Opus 5.5 (Max) 1807.3
67th percentile · #18 of 52 open · #65 of 138 all
#18 of 52
#65 of 138
Arena
Arena · Oct 2026

MiMo V2.5 Pro math benchmarks

Symbolic and quantitative problem-solving, from competition math to multi-step calculation.

ProofBench
ProofBench evaluates a model's ability to write and check rigorous mathematical proofs, rather than just produce a final numeric answer. It probes step-by-step logical correctness, which is a harder and more distinct skill than solving for a number.
22.0%
22.0%
median 17.0%open leader 87.0%
Overall best: Claude Fable 5.1 Max 100.0%
60th percentile · #11 of 26 open · #49 of 82 all
#11 of 26
#49 of 82
Lab-reported
Epoch AI

MiMo V2.5 Pro instruction following benchmarks

How precisely a model obeys explicit, verifiable instructions — format, length, and content constraints — and, more broadly, how much people prefer its responses head-to-head in blind human voting.

LMArena Text
Arena (arena.ai, formerly LMArena / Chatbot Arena) ranks models by millions of blind, head-to-head human votes on real prompts. The Arena Score shown here is a Bradley-Terry rating computed with Arena's default style control, so higher means people preferred that model's answers more often — it's a relative rating, not a percentage, so compare it only to other models on this board.
1467.3
1467.3
median 1321.5open leader 1487.1
Overall best: Gemini 4 Argon (High) 1523.9
97th percentile · #7 of 187 open · #49 of 405 all
#7 of 187
#49 of 405
Arena
Arena · Oct 2026

Score vs VRAM in the ≤640 GB tier

MiMo V2.5 Pro
Other models
Best Score for its VRAM
x: VRAM (GB, log scale) · y: llmrun Score

#14 of 96 that fit ≤640 GB

Each dot is a model at the best quant that fits ≤640 GB. The line joins the models with the best Score for their VRAM; a dot below it is beaten by a model that needs no more memory.

Full ≤640 GB ranking →

MiMo V2.5 Pro vs its neighbours

Same boards for the nearest open models per llmrun Score. The highest score in each row is marked.

Board	MiMo V2.5 Prothis model	Qwen3.8 27Bbest in classQ4_K_M · 18 GB	GLM 5.1nearest by ScoreQ4_K_M · 454 GB	DeepSeek V4 Pronearest by ScoreQ4_K_M · 960 GB	Inklingnearest by ScoreQ4_K_M · 575 GB
llmrun Score	116.4	122.1	117.8	116.5	116.0
CritPt	4.0%	5.4%	4.6%	12.9%	5.4%
DTBench	84.5%	88.0%	—	90.7%	87.5%
LMCA	29.5%	41.4%	—	41.2%	37.6%
SciCode	50.2%	46.6%	44.8%	50.0%	47.0%
ALE-Bench	899.8	—	887.1	1006.1	946.0
ProofBench	22.0%	16.0%	22.2%	16.0%	0.0%
LMArena Text	1467.3	1437.8	1465.2	1457.8	1440.9
Compare head to head		Compare	Compare	Compare	Compare

Scores aggregated from public benchmark sources (Epoch AI: 6 boards, Arena: 2 boards) — each linked from the benchmark pages. llmrun does not run these benchmarks. Percentiles and ranks are among open models in the llmrun catalogue; the median and leader marks refer to the same field.