Up to four models
MiMo-V2.6-Flash vs Kimi K3
On the index (GPT-4 = 100), MiMo-V2.6-Flash scores 168 and Kimi K3 168. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
MiMo-V2.6-Flash
168149–186
WeightsOpen
ReleasedSep 2026
Price in / out$0.14 / $0.28
Context1.05M
Core tests taken2 of 9
Kimi K3
168152–183
WeightsOpen
ReleasedJul 2026
Price in / out$0.19 / $16
Context1.05M
Core tests taken6 of 9
MiMo-V2.6-Flash and Kimi K3 are too close to call (0 points apart, within the uncertainty), ahead on 2 of 2 shared core tests, and costs 24× less.
Test by testthird-party results
MiMo-V2.6-FlashKimi K3
Every result as a table
| Test | MiMo-V2.6-Flash | Kimi K3 |
|---|---|---|
| Maths | ||
| ProofBench v1.1 | 63.0% | 87.0% |
| Coding | ||
| Terminal-Bench 4.0 | 24.2% | 17.2% |
| Code Migration | 40.9% | 16.1% |
| CyberBench v1.1 | 75.4% | 75.2% |
| IOI | 47.7% | 48.9% |
| ProgramBench | 0.5% | 2.0% |
| Vibe Code Bench v1.1 | 79.0% | 85.0% |
| Agents | ||
| BioMysteryBench | 69.3% | 72.6% |
| Terminal-Bench-Science | 4.3% | 1.4% |
| Professional work | ||
| Finance Agent (v2) | 56.3% | 53.1% |
| EMB | 65.5% | 66.7% |
| HLAB-Vals-1.1 | 13.3% | 12.9% |
| Legal Research Bench | 38.0% | 46.2% |
| MedCode | 41.1% | 49.4% |
| MedScribe | 85.3% | 88.0% |
| Public Benefits Bench v1.1 | 67.6% | 68.2% |
| SAGE | 83.3% | 86.8% |
| Tax Agent Bench | 28.5% | 33.4% |
Tests at least two of these models have taken. Bold rows are the core tests.