LLMs TigerLive LLM benchmarks

mimo-v2.5-pro: benchmark scores, price and ranking

mimo-v2.5-pro from Xiaomi Corp ranks #259 of 743 LLMs on the LLMs Tiger aggregate benchmark score (49.5), based on 12 benchmark results captured up to 2026-10-13.

Organisation
Xiaomi Corp
Type
Open-weight / local model
Open weights
Yes
Released
2026-04-22 (source: openrouter)
Cheapest API price
$0.435 in / $0.870 out per million tokens (via openrouter, observed 2026-10-13T19:47Z)
Aggregate rank
#259 of 743 (score 49.5)

Relative to other models, mimo-v2.5-pro is strongest on GSM8K (#1 of 17) and weakest on LMArena Text Elo (#28 of 28).

Across the 11 benchmarks where at least ten models are scored, mimo-v2.5-pro is in the top 10% on 1 and in the top half on 9.

Its cheapest listed input price, $0.435 per million tokens, is lower than that of 53% of the 427 models with a listed price.

mimo-v2.5-pro benchmark results

mimo-v2.5-pro benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
GSM8K99.6%#1 of 17HuggingFace GSM8K leaderboard2026-10-13
DTBench84.5%#39 of 146Epoch AI Benchmarking Hub2026-10-13
MMLU-Pro68.5%#44 of 136HuggingFace MMLU-Pro leaderboard2026-10-13
SciCode50.2%#35 of 106Epoch AI Benchmarking Hub2026-10-13
ALE-Bench900#29 of 79Epoch AI Benchmarking Hub2026-10-13
WebDev Arena1,475#26 of 67Epoch AI Benchmarking Hub2026-10-13
CritPt4.0%#50 of 115Epoch AI Benchmarking Hub2026-10-13
ProofBench22.0%#22 of 46Epoch AI Benchmarking Hub2026-10-13
LMCA29.5%#52 of 108Epoch AI Benchmarking Hub2026-10-13
GPQA Diamond66.7%#151 of 275HuggingFace GPQA Diamond leaderboard2026-10-13
SWE-rebench46.5%#4 of 7SWE-rebench leaderboard2026-10-13
LMArena Text Elo1,468#28 of 28LMArena2026-10-05

About mimo-v2.5-pro

mimo-v2.5-pro is an open-weight model that can run on your own hardware from Xiaomi Corp. It was released on 2026-04-22, according to the date OpenRouter first listed it (approximate). The cheapest API listing we track charges $0.435 per million input tokens and $0.870 per million output tokens, through openrouter. We hold 12 benchmark results for it across 4 areas, and 1 is in the top tenth of its board.

Where mimo-v2.5-pro stands

On our aggregate score, mimo-v2.5-pro is #259 of 743 ranked models, around the middle of the table. Its score is 49.5; the scale is centred on 50, so it sits 0.5 points below the typical model. That beats 65% of the ranked field. Among the 5 ranked Xiaomi Corp models it is #3. The next one up is MiMo-V2.6 Flash at 51.7. The next one down is mimo-v2.5 at 46.7.

mimo-v2.5-pro, benchmark by benchmark

GSM8K is where mimo-v2.5-pro does best: 99.6%, first of 17. The next result is Llama 3.1-405B at 96.8%, 2.8 points behind. It covers grade-school math word problems. The board median is 86.9%, so mimo-v2.5-pro is 12.7 points above it. Source: HuggingFace GSM8K leaderboard, captured 2026-10-13.

On DTBench, mimo-v2.5-pro reaches 84.5%. It ranks #39 of 146, ahead of 73% of the field, and 14.4 points behind Claude Opus 5.5 (max) at 98.9%. The results either side are o3 (high) (84.8%) and Qwen3.7 Plus (84.0%). It covers dynamic tool-use benchmark. The board median is 68.8%, so mimo-v2.5-pro is 15.7 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

mimo-v2.5-pro manages 68.5% on MMLU-Pro, good for #44 of 136. Darwin-180B-RSI leads at 88.1%, 19.6 points further up, and 68% of models sit below. The results either side are Qwen2.5-32B (69.2%) and qwen3-235b-a22b (68.2%). It covers 57-subject knowledge + reasoning. The board median is 55.3%, so mimo-v2.5-pro is 13.2 points above it. Source: HuggingFace MMLU-Pro leaderboard, captured 2026-10-13.

SciCode puts mimo-v2.5-pro at 50.2%, #35 of 106. That beats 67% of the models tested, but 16.7 points short of Claude Opus 5.5 (max) (66.9%). The results either side are grok-build-0.1 (50.2%) and DeepSeek v4 Pro (max) (50.0%). It covers scientific research code. The board median is 44.3%, so mimo-v2.5-pro is 5.9 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On ALE-Bench, mimo-v2.5-pro reaches 900. It ranks #29 of 79, ahead of 63% of the field, and 2,052 points behind GPT-6 Astra (max) at 2,951. The results either side are Gemini 3.7 Flash (high) (904) and GLM-5.1 (887). It covers AtCoder heuristic-optimisation contests (performance rating). The board median is 764, so mimo-v2.5-pro is 136 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

mimo-v2.5-pro manages 1,475 on WebDev Arena, good for #26 of 67. Claude Opus 5.5 (max) leads at 1,820, 345 points further up, and 61% of models sit below. The results either side are Qwen 3.6 Max (Preview) (1,479) and Kimi K2.7 Code (1,473). It covers building web apps, crowd-voted (Elo). The board median is 1,433, so mimo-v2.5-pro is 41 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

ProofBench puts mimo-v2.5-pro at 22.0%, #22 of 46. That beats 52% of the models tested, but 78.0 points short of Claude Fable 5.1 (max) (100.0%). The results either side are GLM-5.1 (22.2%) and GLM-5.3-Flash (max) (21.0%). It covers writing mathematical proofs. The board median is 21.0%, so mimo-v2.5-pro is 1.0 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On LMCA, mimo-v2.5-pro reaches 29.5%. It ranks #52 of 108, ahead of 52% of the field, and 38.7 points behind Claude Opus 5.5 (max) at 68.2%. The results either side are Gemma 4 26B A4B (29.7%) and qwen3.5-35b-a3b (29.5%). It covers language-model capability assessment. The board median is 29.2%, so mimo-v2.5-pro is 0.3 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

mimo-v2.5-pro lands below the middle on GPQA Diamond: 66.7%, #151 of 275, with 55% of the field scoring higher. GPT-6 Astra (max) leads at 95.8%; the closest result above is GPT-4.1 at 66.9%. It covers expert-level science questions. The board median is 71.7%, so mimo-v2.5-pro is 5.0 points below it. Source: HuggingFace GPQA Diamond leaderboard, captured 2026-10-13.

LMArena Text Elo is a weak spot. mimo-v2.5-pro scores 1,468, #28 of 28, behind 96% of the models measured. The leader, gemini-4-argon-high, reaches 1,525. It covers head-to-head chat preference (crowd votes). The board median is 1,478, so mimo-v2.5-pro is 10 points below it. Source: LMArena, captured 2026-10-05.

The full table above also lists the other 2 results, including boards with fewer than ten models, where a rank says little.

Strengths and weaknesses

Software engineering: solid, ahead of 64% of the field on average across 3 benchmarks. Its best board here is SciCode (#35 of 106) and its weakest is WebDev Arena (#26 of 67). Reasoning, maths and science: solid, ahead of 55% of the field on average across 4 benchmarks. Its best board here is MMLU-Pro (#44 of 136) and its weakest is GPQA Diamond (#151 of 275). General knowledge and chat: uneven, ahead of 49% of the field on average across 3 benchmarks. Its best board here is GSM8K (#1 of 17) and its weakest is LMArena Text Elo (#28 of 28).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so mimo-v2.5-pro costs less than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4.1 Flash (high) ($0.010, 48.9), Qwen3.7 Flash ($0.030, 50.5). Similar scores cost more on Kimi K2.5 ($0.450) and Inkling Small (xhigh) ($0.450).

Release history

The Xiaomi Corp model released before it was mimo-v2.5 on 2026-04-22, which scores 46.7. The next one was MiMo-V2.6 Flash on 2026-09-21, at 51.7.

Reading these numbers

The 12 scores on this page come from 6 sources: Epoch AI Benchmarking Hub (7), HuggingFace GPQA Diamond leaderboard (1), HuggingFace GSM8K leaderboard (1), LMArena (1). No official leaderboard supplied any of them, so treat them as third-party measurements.

Models ranked nearby

More from Xiaomi Corp

Compare it on the live leaderboard → · How the aggregate score works