mimo-v2.5: benchmark scores, price and ranking
mimo-v2.5 from Xiaomi Corp ranks #494 of 743 LLMs on the LLMs Tiger aggregate benchmark score (46.7), based on 5 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
- $0.140 in / $0.280 out per million tokens (via openrouter, observed 2026-10-13T20:47Z)
- Aggregate rank
- #494 of 743 (score 46.7)
Relative to other models, mimo-v2.5 is strongest on CritPt (#51 of 115) and weakest on ALE-Bench (#58 of 79).
Across the 5 benchmarks where at least ten models are scored, mimo-v2.5 is in the top 10% on 0 and in the top half on 2.
Its cheapest listed input price, $0.140 per million tokens, is lower than that of 75% of the 427 models with a listed price.
mimo-v2.5 benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| CritPt | 3.7% | #51 of 115 | Epoch AI Benchmarking Hub | 2026-10-13 |
| WebDev Arena | 1,437 | #30 of 67 | Epoch AI Benchmarking Hub | 2026-10-13 |
| SciCode | 43.1% | #56 of 106 | Epoch AI Benchmarking Hub | 2026-10-13 |
| ProofBench | 16.0% | #31 of 46 | Epoch AI Benchmarking Hub | 2026-10-13 |
| ALE-Bench | 514 | #58 of 79 | Epoch AI Benchmarking Hub | 2026-10-13 |
About mimo-v2.5
mimo-v2.5 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 $0.140 per million input tokens and $0.280 per million output tokens, through openrouter. We hold 5 benchmark results for it across 2 areas, and none is in the top tenth of its board.
Where mimo-v2.5 stands
On our aggregate score, mimo-v2.5 is #494 of 743 ranked models, in the lower half. Its score is 46.7; the scale is centred on 50, so it sits 3.3 points below the typical model. That beats 34% of the ranked field. Among the 5 ranked Xiaomi Corp models it is #4. The next one up is mimo-v2.5-pro at 49.5. The next one down is mimo-v2-pro at 46.6.
mimo-v2.5, benchmark by benchmark
CritPt puts mimo-v2.5 at 3.7%, #51 of 115. That beats 56% of the models tested, but 28.6 points short of GPT-5.6 Sol (max) (32.3%). The results either side are mimo-v2.5-pro (4.0%) and MiniMax-M3 (3.7%). It covers unpublished research-level physics problems. The board median is 2.9%, so mimo-v2.5 is 0.9 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
On WebDev Arena, mimo-v2.5 reaches 1,437. It ranks #30 of 67, ahead of 55% of the field, and 383 points behind Claude Opus 5.5 (max) at 1,820. The results either side are Gemini 3 Pro Preview (1,439) and Kimi K2.5 (1,436). It covers building web apps, crowd-voted (Elo). The board median is 1,433, so mimo-v2.5 is 4 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
mimo-v2.5 lands below the middle on SciCode: 43.1%, #56 of 106, with 52% of the field scoring higher. Claude Opus 5.5 (max) leads at 66.9%; the closest result above is Gemma 4 31B IT at 43.4%. It covers scientific research code. The board median is 44.3%, so mimo-v2.5 is 1.3 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
mimo-v2.5 lands below the middle on ProofBench: 16.0%, #31 of 46, with 65% of the field scoring higher. Claude Fable 5.1 (max) leads at 100.0%; the closest result above is Kimi K2.6 at 16.0%. It covers writing mathematical proofs. The board median is 21.0%, so mimo-v2.5 is 5.0 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
mimo-v2.5 lands below the middle on ALE-Bench: 514, #58 of 79, with 72% of the field scoring higher. GPT-6 Astra (max) leads at 2,951; the closest result above is GPT-4.1 at 558. It covers AtCoder heuristic-optimisation contests (performance rating). The board median is 764, so mimo-v2.5 is 250 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Strengths and weaknesses
Software engineering: uneven, ahead of 43% of the field on average across 3 benchmarks. Its best board here is WebDev Arena (#30 of 67) and its weakest is ALE-Bench (#58 of 79). Reasoning, maths and science: uneven, ahead of 44% of the field on average across 2 benchmarks. Its best board here is CritPt (#51 of 115) and its weakest is ProofBench (#31 of 46).
Price and value
The median input price across the 427 priced models is $0.500 per million tokens, so mimo-v2.5 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), Qwen2.5-Coder-3B-Instruct ($0.010, 46.5). Similar scores cost more on DeepSeek-R1-Distill-Qwen-32B ($0.150) and Magistral Small ($0.150).
Release history
The next one was mimo-v2.5-pro on 2026-04-22, at 49.5.
Reading these numbers
The 5 scores on this page come from 1 source: Epoch AI Benchmarking Hub (5). No official leaderboard supplied any of them, so treat them as third-party measurements. With 5 results the aggregate rank can move a lot when one new score arrives.
Models ranked nearby
- #491 Llama 4 Maverick Instruct (46.7)
- #492 internlm2-chat-20b (46.7)
- #493 mistral-small-2402 (46.7)
- #495 Claude Sonnet 3.5 (46.7)
- #496 MCTS Refine 7B (46.7)
- #497 Composer 2.5 (46.6)
More from Xiaomi Corp
- mimo-v2.6-pro (#19, 54.7)
- MiMo-V2.6 Flash (#100, 51.7)
- mimo-v2.5-pro (#259, 49.5)
- mimo-v2-pro (#502, 46.6)
Compare it on the live leaderboard → · How the aggregate score works