Vellum - LiveCodeBench: leaderboard
Vellum's independently-run LiveCodeBench evaluation. Continuously updated coding problems to avoid data contamination.
Metric: Pass@1 (%). Source: www.vellum.ai. Saturation forecast: Estimated already saturated. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4 Pro | 93.5 | #102 |
| 2 | DeepSeek V4 Flash | 91.6 | #137 |
| 3 | Kimi K2.5 | 85 | #144 |
| 4 | Kimi K2 (Thinking) | 83.1 | #243 (Kimi K2) |
| 5 | Gemini 3 Pro | 79.7 | #80 |
| 6 | Grok 3 | 79.4 | #304 |
| 7 | Grok 4 | 79 | #181 |
| 8 | Claude Opus 4.6 | 76 | #63 |
| 9 | O3 Mini | 74.1 | #272 |
| 10 | Claude Sonnet 4.6 | 72.4 | #90 |
| 11 | Gemini 2.5 Pro | 69 | #152 |
| 12 | GPT-OSS-120B | 69 | #341 |
| 13 | GPT-OSS-20B | 69 | #518 |
| 14 | DeepSeek R1 | 64.3 | #251 |
| 15 | Gemini 2.5 Flash | 63.5 | #245 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vellum-livecodebench · How It Works · Data refreshed daily, snapshot 2026-10-20.