MiMo V2.6 Pro RL — Benchmarks
Benchmark scores for MiMo V2.6 Pro RL aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.
Composite Score
134.9
90% range 131–138 · #1 of 100 open models · from 5 benchmarks
See it in the open-model ranking →How the Composite Score is built
Coding
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| LMArena WebDev | 1629.0 | #5 / 50 | #19 / 135 |
| SciCode | 60.9% | #1 / 65 | #8 / 200 |
| ALE-Bench | 1158.1 | #5 / 42 | #41 / 118 |
Instruction Following
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| LMArena Text | 1480.4 | #2 / 186 | #26 / 403 |
Math
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| ProofBench | 70.0% | #3 / 26 | #15 / 79 |
Reasoning
| Benchmark | Score | Open rank | All models |
|---|---|---|---|
| CritPt | 26.6% | #1 / 65 | #35 / 212 |
| APEX-Agents | 59.5% | #1 / 16 | #14 / 52 |
Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.