New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Codeforces Elo benchmark
AI model leaderboard for the Codeforces Elo benchmark. Compare how large language models score on Codeforces Elo, see the full ranking, and understand what this AI benchmark measures. Gemini 3 Pro currently leads with 2,708. Codeforces competitive programming rating estimated from solving rates.
Leaderboard
| # | Model | Organization | Score | Variant | Source |
|---|---|---|---|---|---|
| #1 | Gemini 3 Pro | Google DeepMind | 2,708 | rating-cited-deepseek-v3-2 | official ↗ |
| #2 | gpt-oss-120b | OpenAI | 2,622 | high-with-tools-elo | official ↗ |
| #3 | GPT-5 | OpenAI | 2,537 | rating-cited-deepseek-v3-2 | official ↗ |
| #4 | gpt-oss-20b | OpenAI | 2,516 | high-with-tools-elo | official ↗ |
| #5 | gpt-oss-120b | OpenAI | 2,463 | high-no-tools-elo | official ↗ |
| #6 | gpt-oss-120b | OpenAI | 2,463 | high reasoning, no tools | official ↗ |
| #7 | DeepSeek-V3.2 | DeepSeek | 2,386 | thinking-rating | official ↗ |
| #8 | gpt-oss-20b | OpenAI | 2,230 | high-no-tools-elo | official ↗ |
| #9 | gpt-oss-20b | OpenAI | 2,230 | high reasoning, no tools | official ↗ |
| #10 | o3-mini | OpenAI | 2,130 | high-rating-cited-phi4r | official ↗ |
| #11 | DeepSeek-V3.2-Exp | DeepSeek | 2,121 | from-hf-readme | official ↗ |
| #12 | Ring-1T | Ant Group | 2,088 | thinking-rating | official ↗ |
| #13 | DeepSeek-V3.1-Terminus | DeepSeek | 2,073 | thinking-rating-cited-ring-1t | official ↗ |
| #14 | Qwen3-235B-A22B | Qwen | 2,056 | thinking-rating | official ↗ |
| #15 | Qwen3-235B-A22B-Thinking (Jul 2025) | Qwen | 2,055 | thinking-rating-cited-ring-1t | official ↗ |
| #16 | DeepSeek-R1 | DeepSeek | 2,029 | rating-cited-phi4r | official ↗ |
| #17 | Qwen3-32B | Qwen | 1,977 | thinking-rating | official ↗ |
| #18 | Qwen3-30B-A3B | Qwen | 1,974 | thinking-rating | official ↗ |
| #19 | GPT-5 | OpenAI | 1,918 | thinking-high-rating-cited-ring-1t | official ↗ |
| #20 | Ling-1T | Ant Group | 1,901 | rating | official ↗ |
| #21 | o1 | OpenAI | 1,891 | rating-cited-phi4r | official ↗ |
| #22 | gemini-2.5-pro | Google DeepMind | 1,837 | rating-cited-ring-1t | official ↗ |
| #23 | Qwen3-8B | Qwen | 1,785 | thinking-rating | official ↗ |
| #24 | Qwen3-14B | Qwen | 1,766 | thinking-rating | official ↗ |
| #25 | Phi-4-Reasoning | Microsoft | 1,736 | rating | official ↗ |
| #26 | Phi-4-Reasoning-plus | Microsoft | 1,723 | rating | official ↗ |
| #27 | gemini-2.5-pro | Google DeepMind | 1,675 | lowthink-rating-cited-ling-1t | official ↗ |
| #28 | Qwen3-4B | Qwen | 1,671 | thinking-rating | official ↗ |
| #29 | o1-mini | OpenAI | 1,650 | rating-cited-phi4r | official ↗ |
| #30 | DeepSeek-R1-Distill-Llama-70B | DeepSeek | 1,633 | rating-cited-phi4r | official ↗ |
| #31 | DeepSeek-V3.1-Terminus | DeepSeek | 1,582 | non-thinking-rating-cited-ling-1t | official ↗ |
| #32 | Kimi K2 | Moonshot | 1,574 | instruct-0905-rating-cited-ling-1t | official ↗ |
| #33 | Claude Sonnet 4.5 | Anthropic | 1,480 | rating-cited-deepseek-v3-2 | official ↗ |
| #34 | Qwen3-235B-A22B | Qwen | 1,387 | non-thinking-rating | official ↗ |
| #35 | Qwen3-32B | Qwen | 1,353 | non-thinking-rating | official ↗ |
| #36 | Qwen3-30B-A3B | Qwen | 1,267 | non-thinking-rating | official ↗ |
| #37 | Qwen3-14B | Qwen | 1,200 | non-thinking-rating | official ↗ |
| #38 | GPT-5 | OpenAI | 1,120 | main-rating-cited-ling-1t | official ↗ |
| #39 | Qwen3-8B | Qwen | 1,110 | non-thinking-rating | official ↗ |
| #40 | Qwen3-4B | Qwen | 842 | non-thinking-rating | official ↗ |
Frequently asked questions about Codeforces Elo
What is the Codeforces Elo benchmark?
Codeforces competitive programming rating estimated from solving rates.
How is the Codeforces Elo benchmark scored?
Codeforces Elo is scored using the Elo (pts) metric, where a higher score is better. The maximum achievable score is 4000.000. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.
Which AI model scores highest on Codeforces Elo?
As of the latest reported scores on GenAIList, Gemini 3 Pro achieves the highest result on Codeforces Elo with a score of 2,708.
Is a higher Codeforces Elo score better?
Yes. On Codeforces Elo a higher score indicates better performance, so models near the top of the leaderboard are the strongest.