Up to four models
Gemini 3.1 Pro vs Kimi K2.5
On the index (GPT-4 = 100), Gemini 3.1 Pro scores 161 and Kimi K2.5 145. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
161146–177
WeightsClosed
ReleasedFeb 2026
Price in / out$2.0 / $12
Context1.05M
Core tests taken9 of 9
Kimi K2.5
145130–161
WeightsOpen
ReleasedJan 2026
Price in / out$0.50 / $2.5
Context262K
Core tests taken5 of 9
Gemini 3.1 Pro leads Kimi K2.5 by 16 points, ahead on 5 of 5 shared core tests, at 4.5× the price.
Test by testthird-party results
Every result as a table
| Test | Gemini 3.1 Pro | Kimi K2.5 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 94.4% | 87.6% |
| Humanity's Last Exam | 46.4% | 24.4% |
| GPQA Diamond (Vals AI) | 95.5% | 84.1% |
| MMLU Pro | 91.0% | 85.9% |
| MMMU Pro | 88.2% | 84.3% |
| CL-bench | 20.8% | 19.3% |
| SimpleBench | 79.6% | 46.8% |
| SimpleQA Verified | 73.5% | 34.3% |
| Maths | ||
| AIME-style maths (OTIS mock) | 95.6% | 92.2% |
| FrontierMath (2025 set) | 36.9% | 27.9% |
| FrontierMath tier 4 (2025 set) | 16.7% | 4.2% |
| Coding | ||
| SWE-bench Verified | 75.6% | 73.8% |
| Code Migration | 17.3% | 7.0% |
| LiveCodeBench | 88.5% | 83.9% |
| SWE-bench | 78.8% | 70.0% |
| Vibe Code Bench v1.1 | 32.0% | 17.5% |
| WeirdML | 72.1% | 45.6% |
| Professional work | ||
| Finance Agent (v2) | 43.0% | 35.8% |
| CorpFin v2 | 64.5% | 68.3% |
| EMB | 52.6% | 28.5% |
| Legal Research Bench | 20.7% | 15.9% |
| MedCode | 59.1% | 39.3% |
| MedScribe | 76.1% | 76.4% |
| MortgageTax | 69.4% | 66.5% |
| TaxEval v2 | 72.9% | 74.2% |
| Novel problems | ||
| ARC-AGI-2 | 77.1% | 11.8% |
| ARC-AGI-1 | 98.0% | 65.3% |
Task length · METR
46 sec
Tests at least two of these models have taken. Bold rows are the core tests.