LiveCodeBench Pro Medium: leaderboard
LiveCodeBench Pro medium split measures pass@1 on mid-difficulty live competitive-programming problems.
Metric: Pass@1 (%). Source: web.archive.org. Saturation forecast: Around January 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (High) | 52.11 | #110 (GPT-5.2) |
| 2 | GPT-5 (High) | 40.85 | #98 (GPT-5) |
| 3 | O4 Mini (2025-04-16) (High) | 29.23 | #179 (O4 Mini (2025-04-16)) |
| 4 | Qwen 3 Next 80B A3B (Thinking) | 7.04 | #320 (Qwen 3 Next 80B A3B) |
| 5 | DeepSeek R1 0528 | 1.41 | #222 |
| 6 | Claude Sonnet 4.5 (Thinking) | 0 | #142 (Claude Sonnet 4.5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=livecodebench-pro-medium · How It Works · Data refreshed daily, snapshot 2026-10-20.