Up to four models
DeepSeek V4.1 Flash 1017 vs Kimi K3
On the index (GPT-4 = 100), DeepSeek V4.1 Flash 1017 scores 176 and Kimi K3 173. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
DeepSeek V4.1 Flash 1017176160–192
WeightsOpen
ReleasedOct 2026
Price in / out$0.30 / $1.2
Context1.05M
Core tests taken5 of 9
Kimi K3
173156–189
WeightsOpen
ReleasedJul 2026
Price in / out$0.20 / $20
Context1.05M
Core tests taken6 of 9
DeepSeek V4.1 Flash 1017 and Kimi K3 are too close to call (3 points apart, within the uncertainty), ahead on 4 of 5 shared core tests, and costs 9.8× less.
Test by testthird-party results
DeepSeek V4.1 Flash 1017Kimi K3Every result as a table
| Test | DeepSeek V4.1 Flash 1017 | Kimi K3 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 93.2% | 93.1% |
| SimpleQA Verified | 42.5% | 50.6% |
| Maths | ||
| FrontierMath (tiers 1-3) | 71.2% | 72.2% |
| AIME-style maths (OTIS mock) | 99.4% | 97.2% |
| Chess puzzles | 46.0% | 37.0% |
| FrontierMath (tier 4) | 46.3% | 39.0% |
| Coding | ||
| Terminal-Bench 4.0 | 31.8% | 17.2% |
| Code Migration | 44.2% | 16.1% |
| Vibe Code Bench v1.1 | 87.7% | 85.0% |
| Agents | ||
| Mystery game puzzles | 49.0% | 26.0% |
| Professional work | ||
| Finance Agent (v2) | 57.0% | 53.1% |
| EMB | 61.8% | 66.7% |
| HLAB-Vals-1.1 | 14.2% | 12.9% |
| Legal Research Bench | 45.2% | 46.2% |
| Tax Agent Bench | 44.4% | 33.4% |
| Novel problems | ||
| ARC-AGI-2 | 78.1% | 60.4% |
| ARC-AGI-1 | 94.5% | 94.5% |
Tests at least two of these models have taken. Bold rows are the core tests.
