LiveBench
ReportedQuestions in mathematics, coding, reasoning, data analysis, language and instruction following, refreshed regularly so that they are less likely to be in training data.
- Kind
- Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
- Link
- https://livebench.ai/#/
- Score shown
- Epoch's column "Global average", as a percentage, higher is better. The source gives no error bars.
- Results
- 54 in all: 22 for open-weight models, 28 for closed ones, 4 where Epoch does not say. 19 linked to a model page here.
Best open-weight result: Qwen/QwQ-32B, 72.0%, number 6 of 54 results.
| # | Model | Weights | Setting | Score | Date | Source |
|---|---|---|---|---|---|---|
| 1 | gemini-2.5-pro-exp-03-25Google DeepMind | closed | default | 82.3% | not dated | LiveBench Leaderboard |
| 2 | gpt-5.1-2025-11-13_highOpenAI | closed | effort high | 78.8% | not dated | LiveBench Leaderboard |
| 3 | claude-3-7-sonnet-20250219Anthropic | closed | default | 76.1% | not dated | LiveBench Leaderboard |
| 4 | o3-mini-2025-01-31_highOpenAI | closed | effort high | 75.9% | not dated | LiveBench Leaderboard |
| 5 | o1-2024-12-17_highOpenAI | closed | effort high | 75.7% | not dated | LiveBench Leaderboard |
| 6 | Qwen/QwQ-32BEpoch: QwQ-32B, Alibaba | open | default | 72.0% | not dated | LiveBench Leaderboard |
| 7 | deepseek-ai/DeepSeek-R1Epoch: DeepSeek-R1, DeepSeek | open | default | 71.6% | not dated | LiveBench Leaderboard |
| 8 | o3-mini-2025-01-31_mediumOpenAI | closed | effort medium | 70.0% | not dated | LiveBench Leaderboard |
| 9 | gpt-4.5-preview-2025-02-27OpenAI | closed | default | 69.0% | not dated | LiveBench Leaderboard |
| 10 | gemini-2.0-flash-thinking-exp-01-21Google DeepMind,Google | closed | default | 66.9% | not dated | LiveBench Leaderboard |
| 11 | deepseek-ai/DeepSeek-V3-0324Epoch: DeepSeek-V3-0324, DeepSeek | open | default | 66.9% | not dated | LiveBench Leaderboard |
| 12 | claude-3-7-sonnet-20250219Anthropic | closed | default | 65.6% | not dated | LiveBench Leaderboard |
| 13 | gemini-2.0-pro-exp-02-05Google DeepMind | closed | default | 65.1% | not dated | LiveBench Leaderboard |
| 14 | gemini-exp-1206Google DeepMind,Google | closed | default | 64.1% | not dated | LiveBench Leaderboard |
| 15 | o3-mini-2025-01-31_lowOpenAI | closed | effort low | 62.5% | not dated | LiveBench Leaderboard |
| 16 | qwen2.5-maxAlibaba | closed | default | 62.3% | not dated | LiveBench Leaderboard |
| 17 | gemini-2.0-flash-001Google DeepMind,Google | closed | default | 61.5% | not dated | LiveBench Leaderboard |
| 18 | deepseek-ai/DeepSeek-V3Epoch: DeepSeek-V3, DeepSeek | open | default | 60.5% | not dated | LiveBench Leaderboard |
| 19 | gemini-2.0-flash-expGoogle DeepMind,Google | closed | default | 59.3% | not dated | LiveBench Leaderboard |
| 20 | claude-3-5-sonnet-20241022Anthropic | closed | default | 59.0% | not dated | LiveBench Leaderboard |
| 21 | o1-mini-2024-09-12_mediumOpenAI | closed | effort medium | 57.8% | not dated | LiveBench Leaderboard |
| 22 | gpt-4o-2024-08-06OpenAI | closed | default | 55.3% | not dated | LiveBench Leaderboard |
| 23 | deepseek-ai/DeepSeek-R1-Distill-Llama-70BEpoch: DeepSeek-R1-Distill-Llama-70B, DeepSeek | open | default | 54.5% | not dated | LiveBench Leaderboard |
| 24 | grok-2-1212xAI | closed | default | 54.3% | not dated | LiveBench Leaderboard |
| 25 | gemini-2.0-flash-liteGoogle DeepMind | closed | default | 54.3% | not dated | LiveBench Leaderboard |
| 26 | gemini-2.0-flash-lite-preview-02-05Google DeepMind | closed | default | 53.2% | not dated | LiveBench Leaderboard |
| 27 | Dracarys2-72B-Instruct | not stated | default | 52.6% | not dated | LiveBench Leaderboard |
| 28 | gpt-4o-2024-11-20OpenAI | closed | default | 52.2% | not dated | LiveBench Leaderboard |
| 29 | learnlm-1.5-pro-experimental | not stated | default | 52.2% | not dated | LiveBench Leaderboard |
| 30 | meta-llama/Llama-3.3-70B-InstructEpoch: Llama-3.3-70B-Instruct, Meta AI | open | default | 50.2% | not dated | LiveBench Leaderboard |
| 31 | google/gemma-3-27b-itEpoch: gemma-3-27b-it, Google DeepMind | open | default | 50.0% | not dated | LiveBench Leaderboard |
| 32 | claude-3-opus-20240229Anthropic | closed | default | 49.2% | not dated | LiveBench Leaderboard |
| 33 | mistral-large-2411Mistral AIno page match | open | default | 48.4% | not dated | LiveBench Leaderboard |
| 34 | sonar | not stated | default | 46.9% | not dated | LiveBench Leaderboard |
| 35 | Qwen/Qwen2.5-Coder-32B-InstructEpoch: Qwen2.5-Coder-32B-Instruct, Alibaba | open | default | 46.2% | not dated | LiveBench Leaderboard |
| 36 | Dracarys2-Llama-3.1-70B-Instruct | not stated | default | 46.2% | not dated | LiveBench Leaderboard |
| 37 | deepseek-ai/DeepSeek-R1-Distill-Qwen-32BEpoch: DeepSeek-R1-Distill-Qwen-32B, DeepSeek | open | default | 45.6% | not dated | LiveBench Leaderboard |
| 38 | mistral-small-2503Mistral AIno page match | open | default | 44.0% | not dated | LiveBench Leaderboard |
| 39 | amazon.nova-pro-v1:0Amazon | closed | default | 43.5% | not dated | LiveBench Leaderboard |
| 40 | claude-3-5-haiku-20241022Anthropic | closed | default | 43.5% | not dated | LiveBench Leaderboard |
| 41 | mistral-small-2501Mistral AIno page match | open | default | 42.5% | not dated | LiveBench Leaderboard |
| 42 | microsoft/phi-4Epoch: phi-4, Microsoft Research | open | default | 41.6% | not dated | LiveBench Leaderboard |
| 43 | gpt-4o-mini-2024-07-18OpenAI | closed | default | 41.3% | not dated | LiveBench Leaderboard |
| 44 | Qwen/QwQ-32B-PreviewEpoch: QwQ-32B-Preview, Alibaba | open | default | 40.3% | not dated | LiveBench Leaderboard |
| 45 | google/gemma-2-27b-itEpoch: gemma-2-27b-it, Google DeepMind | open | default | 38.2% | not dated | LiveBench Leaderboard |
| 46 | amazon.nova-lite-v1:0Amazon | closed | default | 36.4% | not dated | LiveBench Leaderboard |
| 47 | CohereLabs/c4ai-command-r-plus-08-2024Epoch: c4ai-command-r-plus-08-2024, Cohere,Cohere Labs (formerly Cohere for AI) | open | default | 31.8% | not dated | LiveBench Leaderboard |
| 48 | amazon.nova-micro-v1:0Amazon | closed | default | 29.6% | not dated | LiveBench Leaderboard |
| 49 | google/gemma-2-9b-itEpoch: gemma-2-9b-it, Google DeepMind | open | default | 28.7% | not dated | LiveBench Leaderboard |
| 50 | CohereLabs/c4ai-command-r-08-2024Epoch: c4ai-command-r-08-2024 | open | default | 27.5% | not dated | LiveBench Leaderboard |
| 51 | microsoft/Phi-3-small-8k-instructEpoch: Phi-3-small-8k-instruct, Microsoft | open | default | 24.0% | not dated | LiveBench Leaderboard |
| 52 | microsoft/Phi-3-mini-4k-instructEpoch: Phi-3-mini-4k-instruct, Microsoft | open | default | 22.4% | not dated | LiveBench Leaderboard |
| 53 | allenai/OLMo-2-1124-13B-InstructEpoch: OLMo-2-1124-13B-Instruct, Allen Institute for AI,University of Washington,New York University (NYU) | open | default | 22.1% | not dated | LiveBench Leaderboard |
| 54 | microsoft/Phi-3-mini-4k-instructEpoch: Phi-3-mini-4k-instruct, Microsoft | open | default | 22.1% | not dated | LiveBench Leaderboard |
Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method