LiveOIBench: leaderboard
Contamination-free Olympiad benchmark with 403 expert-curated problems from 14 Informatics Olympiads (IOI, USACO, APIO, etc). Models ranked by human percentile against actual contestants.
Metric: Avg Human Percentile. Source: liveoibench.github.io. Saturation forecast: Around January 2027. 58 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 80.18 | #98 |
| 2 | GPT-OSS-120B (High) | 71.85 | #341 (GPT-OSS-120B) |
| 3 | Grok 4 Fast (Reasoning) | 69.8 | #254 (Grok 4 Fast) |
| 4 | Gemini 2.5 Pro | 67.44 | #152 |
| 5 | O3 Mini (High) | 60.86 | #272 (O3 Mini) |
| 6 | GPT-OSS-120B (Medium) | 57.55 | #341 (GPT-OSS-120B) |
| 7 | GPT-OSS-20B (High) | 57.13 | #518 (GPT-OSS-20B) |
| 8 | Gemini 2.5 Flash | 53.75 | #245 |
| 9 | Claude Sonnet 4.5 | 51.39 | #142 |
| 10 | GPT-OSS-20B (Medium) | 51.12 | #518 (GPT-OSS-20B) |
| 11 | GPT-OSS-120B (Low) | 44.07 | #341 (GPT-OSS-120B) |
| 12 | Qwen 3 32B | 41.95 | #443 |
| 13 | DeepSeek R1 | 38.8 | #251 |
| 14 | Qwen 3 30B A3B | 35.2 | #493 |
| 15 | GPT-OSS-20B (Low) | 34.64 | #518 (GPT-OSS-20B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=liveoibench · How It Works · Data refreshed daily, snapshot 2026-10-20.