MiMo-V2.5-Pro
Best configuration ranks #113 of 610 on the BenchLeader Index at 59.0 ±4.5. Last measured 8 Sept 2026. Released 23 Apr 2026.
- Blended price
- $0.653/M
- $0.522 in · $1.04 out
- Output speed
- 35 tok/s
- First answer
- 60 s
- first token 3.26 s
- Context
- 1.0M
- Overall index59
- Reasoning67
- Coding58
- Agents & tools51
- Maths43
- Knowledge58
- Instruction following77
- Human preference62
- Long context66
- Composite62
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 50.5 | #284 | 33 tok/s | 4.60 s | $0.0004 | Agents & tools 65 · Composite 52 · Instruction following 45 · Knowledge 46 · Long context 46 |
| defaultbest | 59.0 | #113 | 35 tok/s | 60 s | $0.0005 | Agents & tools 51 · Coding 58 · Composite 62 · Human preference 62 · Instruction following 77 · Knowledge 58 · Long context 66 · Maths 43 · Reasoning 67 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| LMArena Hard Prompts | – | 1496#32 | LMArena | |
| GPQA Diamond (AA)not in index | 76.2%#231 | 86.6%#109 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 14.8%#197 | 35.7%#77 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | 82.6%#59 | Vals AI |
Coding
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| SciCode | – | 50.2%#57 | SciCode | |
| LMArena Coding | – | 1521#26 | LMArena | |
| LMArena WebDev | – | 1475#46 | LMArena | |
| SciCode (AA)not in index | – | 50.6%#63 | Artificial Analysis | |
| LiveCodeBench | – | 81.3%#61 | Vals AI | |
| SWE-bench (Vals)not in index | – | 74.0%#48 | Vals AI |
Agents & tools
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| LMArena Agent | – | -5.1#31 | LMArena | |
| Terminal-Bench Hard | 35.6%#72 | 43.2%#39 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 72.5%#144 | 94.2%#34 | Artificial Analysis | |
| Terminal-Bench 2.1 (Vals) | – | 57.3%#34 | Vals AI |
Maths
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| ProofBench | – | 22.0%#33 | Vals AI |
Knowledge
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AA-Omniscience | -37.8#249 | 3.3#80 | Artificial Analysis | |
| MMLU-Pro | – | 84.6%#60 | Vals AI | |
| LegalBench | – | 77.1%#101 | Vals AI | |
| CorpFin | – | 61.4%#53 | Vals AI | |
| TaxEval | – | 73.8%#45 | Vals AI |
Instruction following
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| IFBench | 42.7%#230 | 79.9%#9 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| LMArena Text | – | 1468#41 | LMArena | |
| EQ-Bench 4 | – | 1208#14 | EQ-Bench |
Long context
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AA-LCR | 41.7%#282 | 79.7%#59 | Artificial Analysis |
Composite
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AA Intelligence Index | 18.3#191 | 26.4#107 | Artificial Analysis | |
| Vals Indexnot in index | – | 41.0#35 | Vals AI |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| AtlasCloud | 31 tok/s | 5.41 s | $0.435 | $0.870 | 1.0M | fp8 |
| Xiaomi | 29 tok/s | 3.26 s | $0.435 | $0.870 | 1.0M | fp8 |
| DeepInfra | 28 tok/s | 1.81 s | $0.390 | $1.17 | 1.0M | fp8 |
| GMICloud | 26 tok/s | 3.70 s | $0.304 | $0.609 | 1.1M | bf16 |
| DigitalOcean | 25 tok/s | 0.87 s | $0.400 | $1.50 | 262k | – |
| StreamLake | 24 tok/s | 2.47 s | $0.522 | $1.04 | 1M | – |
| NovitaAI | 18 tok/s | 6.54 s | $0.480 | $0.961 | 1.0M | – |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.200 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0005 | $0.0004 | 1.1 min |
| Summarise a 30-page report | 12,000 / 600 | $0.0069 | $0.0040 | 1.3 min |
| Code edit | 6,000 / 1,500 | $0.0047 | $0.0032 | 1.7 min |
| Agentic coding session | 60,000 / 4,000 | $0.035 | $0.021 | 2.9 min |
| Structured extraction | 2,000 / 200 | $0.0013 | $0.0008 | 1.1 min |
See also
Data as of 9 Sept 2026. Compare these configurations.