Up to four models
Kimi K2.7 Code vs MiniMax-M3
On the index (GPT-4 = 100), Kimi K2.7 Code scores 151 and MiniMax-M3 147. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Kimi K2.7 Code
151135–167
WeightsOpen
ReleasedJun 2026
Price in / out$0.70 / $3.75
Context262K
Core tests taken3 of 9
147131–164
WeightsOpen
ReleasedJun 2026
Price in / out$0.30 / $1.2
Context1.05M
Core tests taken4 of 9
Kimi K2.7 Code and MiniMax-M3 are too close to call (3 points apart, within the uncertainty), ahead on 0 of 2 shared core tests, at 2.8× the price.
Test by testthird-party results
Kimi K2.7 Code
MiniMax-M3
Every result as a table
| Test | Kimi K2.7 Code | MiniMax-M3 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 87.9% | 90.9% |
| LiveBench Data Analysis | 62.7% | 76.2% |
| LiveBench IF | 56.3% | 57.5% |
| LiveBench Language | 77.9% | 76.8% |
| LiveBench Reasoning | 82.8% | 74.5% |
| SimpleBench | 57.9% | 45.8% |
| Maths | ||
| AIME-style maths (OTIS mock) | 95.6% | 71.1% |
| LiveBench Mathematics | 79.6% | 77.0% |
| Chess puzzles | 14.0% | 17.0% |
| Surface Evolver Bench | 48.8% | 55.0% |
| Coding | ||
| LiveBench Agentic Coding | 45.7% | 40.7% |
| LiveBench Coding | 74.0% | 68.2% |
| Code Migration | 25.4% | 19.9% |
| LiveCodeBench | 82.0% | 82.2% |
| SWE-bench | 78.2% | 75.0% |
| Vibe Code Bench v1.1 | 47.2% | 47.6% |
| Agents | ||
| APEX-Agents | 37.6% | 37.7% |
| SkillsBench | 50.0% | 51.5% |
Tests at least two of these models have taken. Bold rows are the core tests.