Xentir DeepSeek-V4-Pro vs Kimi K2.5
DeepSeek-V4-Pro OPEN WEIGHTS · DeepSeek, released 2026-04-24
Kimi K2.5 OPEN WEIGHTS · Moonshot, released 2026-01-27
“Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks' [online resource].” · epoch.ai/benchmarks · Licensed CC BY 4.0 · data last verified 2026-09-09 · Methodology · Data Sources
DeepSeek-V4-Pro leads on more of the 6 benchmarks where both are measured — DeepSeek-V4-Pro on 6. The remaining 3 of 9 tracked benchmarks cannot be compared: 2 have a result for only one of the two models, 1 for neither. Nothing on this page is estimated to fill those in.
Head-to-head
Measured on 6 of 9 tracked benchmarks for both models — DeepSeek-V4-Pro leads on 6.
| Benchmark | DeepSeek-V4-Pro | Kimi K2.5 | Leads |
|---|---|---|---|
| Epoch Capabilities Index (ECI) | 149.1 | 148.0 | DeepSeek-V4-Pro |
| SWE-bench Verified | 0.776 (max) | 0.738 | DeepSeek-V4-Pro |
| GPQA Diamond | 0.909 (high) | 0.876 (fireworks) | DeepSeek-V4-Pro |
| MATH Level 5 | Not measured | Not measured | Not measured on both |
| OTIS Mock AIME 2024-2025 | 0.967 (max) | 0.922 (fireworks) | DeepSeek-V4-Pro |
| FrontierMath | Not measured | 0.279 (fireworks) | Comparison unavailable — DeepSeek-V4-Pro has no published result |
| FrontierMath Tier 4 | Not measured | 0.042 (fireworks) | Comparison unavailable — DeepSeek-V4-Pro has no published result |
| SimpleQA Verified | 0.470 (max) | 0.343 | DeepSeek-V4-Pro |
| Chess Puzzles | 0.200 (max) | 0.120 (fireworks) | DeepSeek-V4-Pro |
What remains uncertain
- DeepSeek-V4-Pro has no published result on FrontierMath, FrontierMath Tier 4.
- 1 tracked benchmark has no published result for either model.
A missing result means Xentir has not found one published by the source it cites, not that the model performs badly. See Data Sources.
Full profiles
DeepSeek-V4-Pro → · Kimi K2.5 → · the full AI Model Rankings table