other benchmark · included in ECI
ARC AI2
ARC AI2 benchmark (score column: Challenge score).
Results
45
Random baseline
25.0%
Score ceiling
100%
Released
14 Mar 2018
Score by model release date
Best score by organisation
Leaderboard
| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | DeepSeek V3 Open source DeepSeek | 95.3% | — | 26 Dec 2024 | — | DeepSeek-V3 Technical Report ↗ | |
| 2 | Llama 3.1-405B Open source Meta AI | 95.3% | — | 23 Jul 2024 | — | DeepSeek-V3 Technical Report ↗ | |
| 3 | Qwen2.5-72B Open source Alibaba | 94.5% | — | 19 Sep 2024 | — | DeepSeek-V3 Technical Report ↗ | |
| 4 | DeepSeek-V2 (MoE-236B, May 2024) Open source DeepSeek | 92.2% | — | 7 May 2024 | — | DeepSeek-V3 Technical Report ↗ | |
| 5 | phi-3-medium 14B Open source Microsoft | 91.6% | — | 23 Apr 2024 | — | Phi-3 Technical Report: A H… ↗ | |
| 6 | phi-3-small 7.4B Open source Microsoft | 90.7% | — | 23 Apr 2024 | — | Phi-3 Technical Report: A H… ↗ | |
| 7 | Mixtral 8x7B Open source Mistral AI | 87.3% | — | 11 Dec 2023 | — | Phi-3 Technical Report: A H… ↗ | |
| 8 | Stable Beluga 2 Open source Stability AI | 86.1% | — | 20 Jul 2023 | — | Qwen Technical Report ↗ | |
| 9 | phi-3-mini 3.8B Open source Microsoft | 84.9% | — | 23 Apr 2024 | — | Phi-3 Technical Report: A H… ↗ | |
| 10 | Qwen-14B Open source Alibaba | 84.4% | — | 24 Sep 2023 | — | Qwen Technical Report ↗ | |
| 11 | Llama 3-8B Open source Meta AI | 82.8% | — | 18 Apr 2024 | — | Phi-3 Technical Report: A H… ↗ | |
| 12 | Mistral 7B v0.1 Open source Mistral AI | 78.6% | — | 27 Sep 2023 | — | Phi-3 Technical Report: A H… ↗ | |
| 13 | Gemma 7B Open source Google DeepMind | 78.3% | — | 21 Feb 2024 | — | Phi-3 Technical Report: A H… ↗ | |
| 14 | Llama 2-70B Open source Meta AI | 78.3% | — | 18 Jul 2023 | — | Qwen Technical Report ↗ | |
| 15 | Phi-2 Open source Microsoft | 75.9% | — | 12 Dec 2023 | — | Phi-3 Technical Report: A H… ↗ | |
| 16 | Qwen-7B Open source Alibaba | 75.3% | — | 28 Sep 2023 | — | Qwen Technical Report ↗ | |
| 17 | Qwen2.5-Coder-32B Open source Alibaba | 70.5% | — | 18 Sep 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 18 | LLaMA-65B Open source Meta AI | 69.5% | — | 24 Feb 2023 | — | Qwen Technical Report ↗ | |
| 19 | Falcon-180B Open source Technology Innovation Institute | 67.8% | — | 6 Sep 2023 | — | The Falcon Series of Open L… ↗ | |
| 20 | LLaMA-33B Open source Meta AI | 67.5% | — | 24 Feb 2023 | — | Qwen Technical Report ↗ | |
| 21 | DeepSeek-Coder-V2 236B Open source DeepSeek | 64.3% | — | 17 Jun 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 22 | Falcon-40B Open source Technology Innovation Institute | 61.9% | — | 25 May 2023 | — | Falcon2-11B Technical Report ↗ | |
| 23 | Qwen2.5-Coder (7B) Open source Alibaba | 60.9% | — | 18 Sep 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 24 | Llama 2-13B Open source Meta AI | 60.3% | — | 18 Jul 2023 | — | Qwen Technical Report ↗ | |
| 25 | LLaMA-13B Open source Meta AI | 52.7% | — | 24 Feb 2023 | — | Qwen Technical Report ↗ | |
| 26 | MPT-30B Open source MosaicML | 50.6% | — | 22 Jun 2023 | — | Llama 2: Open Foundation an… ↗ | |
| 27 | Yi 6B Open source 01.AI | 50.3% | — | 22 Nov 2023 | — | Yi: Open Foundation Models … ↗ | |
| 28 | Falcon-7B Open source Technology Innovation Institute | 47.9% | — | 24 Apr 2023 | — | Falcon2-11B Technical Report ↗ | |
| 29 | LLaMA-7B Open source Meta AI | 47.6% | — | 24 Feb 2023 | — | Qwen Technical Report ↗ | |
| 30 | StarCoder 2 15B Open source Hugging Face,ServiceNow,Nvidia,BigCode | 47.2% | — | 20 Feb 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 31 | Llama 2-7B Open source Meta AI | 45.9% | — | 18 Jul 2023 | — | The Falcon Series of Open L… ↗ | |
| 32 | Qwen2.5-Coder (1.5B) Open source Alibaba | 45.2% | — | 18 Sep 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 33 | Phi-1.5 Open source Microsoft | 44.4% | 5 | 11 Sep 2023 | — | Textbooks Are All You Need … ↗ | |
| 34 | MPT-7B Open source MosaicML | 42.6% | — | 5 May 2023 | — | Qwen Technical Report ↗ | |
| 35 | DeepSeek Coder 33B Open source DeepSeek,Peking University | 42.2% | — | 2 Nov 2023 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 36 | Gemma 2B Open source Google DeepMind | 42.1% | — | 21 Feb 2024 | — | Gemma: Open Models Based on… ↗ | |
| 37 | XGen-7B Open source Salesforce | 41.2% | — | 27 Jun 2023 | — | XGen-7B Technical Report ↗ | |
| 38 | Dolly 2.0-12b Open source Databricks | 39.6% | — | 11 Apr 2023 | — | XGen-7B Technical Report ↗ | |
| 39 | StarCoder 2 7B Open source Hugging Face,ServiceNow,Nvidia,BigCode | 38.7% | — | 20 Feb 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 40 | Baichuan2-13B Open source Baichuan | 38.0% | — | 6 Sep 2023 | — | Qwen Technical Report ↗ | |
| 41 | DeepSeek Coder 6.7B Open source DeepSeek,Peking University | 36.4% | — | 2 Nov 2023 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 42 | StarCoder 2 3B Open source Hugging Face,ServiceNow,Nvidia,BigCode | 34.2% | — | 22 Feb 2024 | — | Qwen2.5-Coder Technical Rep… ↗ | |
| 43 | Baichuan 2-7B Open source Baichuan | 32.5% | — | 20 Sep 2023 | — | Qwen Technical Report ↗ | |
| 44 | Cerebras-GPT-13B Open source Cerebras Systems | 32.4% | — | 20 Mar 2023 | — | XGen-7B Technical Report ↗ | |
| 45 | DeepSeek Coder 1.3B Open source DeepSeek,Peking University | 25.4% | — | 2 Nov 2023 | — | Qwen2.5-Coder Technical Rep… ↗ |
Caveats: settings (reasoning effort, agent scaffold) differ between rows and materially affect scores; the best reported setting per model is shown. Source: Epoch AI, CC BY 4.0.