Vellum - LiveCodeBench — leaderboard
Vellum's independently-run LiveCodeBench evaluation. Continuously updated coding problems to avoid data contamination.
Metric: Pass@1 (%). Source: www.vellum.ai. Status: saturation imminent. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V4 Pro | 93.5 |
| 2 | DeepSeek V4 Flash | 91.6 |
| 3 | Kimi K2.5 | 85 |
| 4 | Kimi K2 (Thinking) | 83.1 |
| 5 | Gemini 3 Pro | 79.7 |
| 6 | Grok 3 | 79.4 |
| 7 | Grok 4 | 79 |
| 8 | Claude Opus 4.6 | 76 |
| 9 | O3 Mini | 74.1 |
| 10 | Claude Sonnet 4.6 | 72.4 |
| 11 | Gemini 2.5 Pro | 69 |
| 12 | GPT-OSS-120B | 69 |
| 13 | GPT-OSS-20B | 69 |
| 14 | DeepSeek R1 | 64.3 |
| 15 | Gemini 2.5 Flash | 63.5 |
Interactive version: theaggregate.ai/benchmark?slug=vellum-livecodebench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.