收录成绩109109 个模型或系统
当前第一77.5Claude-Opus-4-5-20251101 (FC)
指标Overall Accuracy各评测子类准确率的未加权平均(%)
数据时效自动同步同步于 10-22 15:32 · 1 个数据源
77.5官方榜单
73.2官方榜单
72.5官方榜单
排名表
显示 109 / 109
| 单轮 | 多轮 | Agentic | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 模型 | 机构 | 开源 | 来源 | 备注 | ||||||||
| 1 | Claude-Opus-4-5-20251101 (FC) | Anthropic | 闭源 | 77.5 | 88.6 | 79.8 | 68.4 | 84.5 | 73.8 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 2 | Claude-Sonnet-4-5-20250929 (FC) | Anthropic | 闭源 | 73.2 | 88.7 | 81.1 | 61.4 | 81.0 | 65.0 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 3 | Gemini-3-Pro-Preview (Prompt) | 闭源 | 72.5 | 90.7 | 83.1 | 60.8 | 80.0 | 61.7 | 官方榜单 | 2026-10-22 | 许可证:Proprietary | |
| 4 | GLM-4.6 (FC thinking) | Zhipu AI | 开源 | 72.4 | 87.6 | 80.9 | 68.0 | 77.5 | 55.7 | 官方榜单 | 2026-10-22 | 许可证:MIT |
| 5 | Grok-4-1-fast-reasoning (FC) | xAI | 闭源 | 69.6 | 88.3 | 78.5 | 58.9 | 82.5 | 54.0 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 6 | Claude-Haiku-4-5-20251001 (FC) | Anthropic | 闭源 | 68.7 | 86.5 | 78.7 | 53.6 | 83.5 | 54.4 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 7 | Gemini-3-Pro-Preview (FC) | 闭源 | 68.1 | 85.8 | 81.7 | 63.1 | 68.5 | 54.8 | 官方榜单 | 2026-10-22 | 许可证:Proprietary | |
| 8 | o3-2025-04-16 (Prompt) | OpenAI | 闭源 | 63.0 | 81.9 | 73.2 | 62.3 | 50.5 | 51.8 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 9 | Grok-4-0709 (Prompt) | xAI | 闭源 | 63.0 | 82.8 | 72.5 | 47.0 | 74.0 | 50.5 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 10 | Grok-4-0709 (FC) | xAI | 闭源 | 61.4 | 85.4 | 75.6 | 33.9 | 82.0 | 55.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 11 | Moonshotai-Kimi-K2-Instruct (FC) | MoonshotAI | 开源 | 59.1 | 81.6 | 78.7 | 50.6 | 66.5 | 29.0 | 官方榜单 | 2026-10-22 | 许可证:modified-mit |
| 12 | Grok-4-1-fast-non-reasoning (FC) | xAI | 闭源 | 58.3 | 88.1 | 77.9 | 46.8 | 75.0 | 26.2 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 13 | Command A Reasoning (FC) | Cohere | 开源 | 57.1 | 86.3 | 78.6 | 50.1 | 55.5 | 28.8 | 官方榜单 | 2026-10-22 | 许可证:CC-BY-NC 4.0 License (w/ Acceptable Use Addendum) |
| 14 | DeepSeek-V3.2-Exp (Prompt + Thinking) | DeepSeek | 开源 | 56.7 | 85.5 | 76.0 | 44.9 | 58.0 | 44.1 | 官方榜单 | 2026-10-22 | 许可证:MIT |
| 15 | Gemini-2.5-Flash (FC) | 闭源 | 56.2 | 85.0 | 74.4 | 36.3 | 59.0 | 41.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary | |
| 16 | GPT-5.2-2025-12-11 (FC) | OpenAI | 闭源 | 55.9 | 81.8 | 70.4 | 28.1 | 75.5 | 45.8 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 17 | GPT-5-mini-2025-08-07 (FC) | OpenAI | 闭源 | 55.5 | 69.8 | 58.6 | 27.5 | 82.0 | 44.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 18 | xLAM-2-32b-fc-r (FC) | Salesforce | 开源 | 54.7 | 89.6 | 75.5 | 69.5 | 25.5 | 20.9 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 19 | DeepSeek-V3.2-Exp (FC) | DeepSeek | 开源 | 54.1 | 34.9 | 53.7 | 37.4 | 69.5 | 54.2 | 官方榜单 | 2026-10-22 | 许可证:MIT |
| 20 | GPT-4.1-2025-04-14 (FC) | OpenAI | 闭源 | 54.0 | 82.8 | 70.0 | 38.9 | 68.0 | 23.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 21 | o4-mini-2025-04-16 (FC) | OpenAI | 闭源 | 53.2 | 37.7 | 66.1 | 41.8 | 75.5 | 34.2 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 22 | xLAM-2-70b-fc-r (FC) | Salesforce | 开源 | 53.1 | 88.4 | 72.2 | 77.4 | 15.0 | 14.4 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 23 | Qwen3-235B-A22B-Instruct-2507 (Prompt) | Qwen | 开源 | 52.1 | 90.3 | 78.7 | 44.6 | 50.5 | 19.4 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 24 | GPT-5-nano-2025-08-07 (FC) | OpenAI | 闭源 | 51.5 | 68.0 | 59.4 | 34.5 | 72.5 | 24.7 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 25 | Nanbeige4-3B-Thinking-2511 (FC) | Nanbeige | 开源 | 51.4 | 81.6 | 79.4 | 51.1 | 21.5 | 36.8 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 26 | Gemini-2.5-Flash (Prompt) | 闭源 | 50.9 | 88.1 | 78.2 | 16.8 | 62.0 | 38.7 | 官方榜单 | 2026-10-22 | 许可证:Proprietary | |
| 27 | GPT-4.1-mini-2025-04-14 (FC) | OpenAI | 闭源 | 50.5 | 83.8 | 68.8 | 34.1 | 57.0 | 26.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 28 | o4-mini-2025-04-16 (Prompt) | OpenAI | 闭源 | 50.3 | 81.3 | 70.8 | 16.6 | 71.5 | 35.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 29 | Qwen3-32B (FC) | Qwen | 开源 | 48.7 | 88.8 | 82.0 | 47.9 | 21.5 | 26.7 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 30 | o3-2025-04-16 (FC) | OpenAI | 闭源 | 48.6 | 40.4 | 66.2 | 14.8 | 77.0 | 47.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 31 | Qwen3-235B-A22B-Instruct-2507 (FC) | Qwen | 开源 | 48.0 | 37.4 | 68.9 | 45.4 | 54.0 | 23.9 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 32 | Nanbeige3.5-Pro-Thinking (FC) | Nanbeige | 开源 | 47.7 | 38.4 | 70.0 | 40.0 | 42.0 | 45.2 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 33 | Qwen3-32B (Prompt) | Qwen | 开源 | 46.8 | 90.3 | 82.0 | 43.3 | 26.0 | 15.7 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 34 | xLAM-2-8b-fc-r (FC) | Salesforce | 开源 | 46.7 | 84.6 | 68.0 | 70.0 | 6.5 | 14.0 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 35 | Command A (FC) | Cohere | 开源 | 46.5 | 87.6 | 78.5 | 29.5 | 46.5 | 16.6 | 官方榜单 | 2026-10-22 | 许可证:CC-BY-NC 4.0 License (w/ Acceptable Use Addendum) |
| 36 | BitAgent-Bounty-8B | Bittensor | 开源 | 46.2 | 81.6 | 93.1 | 62.4 | 0.0 | 1.5 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 37 | Arch-Agent-32B | katanemo | 开源 | 45.4 | 88.9 | 80.7 | 54.3 | 5.0 | 14.6 | 官方榜单 | 2026-10-22 | 许可证:katanemo-research |
| 38 | GPT-5.2-2025-12-11 (Prompt) | OpenAI | 闭源 | 45.3 | 78.3 | 67.1 | 43.8 | 40.5 | 3.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 39 | Qwen3-8B (FC) | Qwen | 开源 | 42.6 | 87.6 | 80.5 | 41.8 | 12.0 | 14.6 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 40 | ToolACE-2-8B (FC) | Huawei Noah & USTC | 开源 | 42.4 | 87.1 | 77.4 | 38.4 | 8.5 | 18.5 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 41 | Qwen3-30B-A3B-Instruct-2507 (FC) | Qwen | 开源 | 41.4 | 85.8 | 77.9 | 30.0 | 22.5 | 17.6 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 42 | xLAM-2-3b-fc-r (FC) | Salesforce | 开源 | 41.2 | 83.0 | 62.9 | 58.4 | 2.5 | 11.4 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 43 | Qwen3-14B (FC) | Qwen | 开源 | 41.0 | 84.9 | 80.0 | 34.8 | 10.0 | 19.6 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 44 | Qwen3-8B (Prompt) | Qwen | 开源 | 40.4 | 88.6 | 80.1 | 33.4 | 13.5 | 13.1 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 45 | GPT-4.1-2025-04-14 (Prompt) | OpenAI | 闭源 | 39.4 | 88.7 | 78.9 | 9.8 | 35.0 | 21.5 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 46 | mistral-large-2411 (FC) | Mistral AI | 闭源 | 38.4 | 84.7 | 81.9 | 14.1 | 28.0 | 24.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 47 | Qwen3-14B (Prompt) | Qwen | 开源 | 37.8 | 89.5 | 79.3 | 26.1 | 10.5 | 11.2 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 48 | Mistral-Medium-2505 | Mistral AI | 闭源 | 37.7 | 85.3 | 66.0 | 9.9 | 39.0 | 21.7 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 49 | Mistral-Medium-2505 (FC) | Mistral AI | 闭源 | 37.6 | 67.4 | 68.0 | 10.8 | 35.0 | 23.0 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 50 | Llama-4-Maverick-17B-128E-Instruct-FP8 (FC) | Meta | 开源 | 37.3 | 88.7 | 73.7 | 20.3 | 28.0 | 18.9 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 4 Community |
| 51 | Mistral-small-2506 (FC) | Mistral AI | 闭源 | 37.1 | 73.6 | 77.3 | 11.5 | 31.0 | 18.1 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 52 | Gemini-2.5-Flash-Lite (FC) | 闭源 | 36.9 | 86.6 | 65.8 | 13.5 | 21.0 | 20.6 | 官方榜单 | 2026-10-22 | 许可证:Proprietary | |
| 53 | Qwen3-30B-A3B-Instruct-2507 (Prompt) | Qwen | 开源 | 36.7 | 88.9 | 78.4 | 23.5 | 17.5 | 9.7 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 54 | Qwen3-4B-Instruct-2507 (FC) | Qwen | 开源 | 35.7 | 87.9 | 76.4 | 22.1 | 3.0 | 17.6 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 55 | Qwen3-4B-Instruct-2507 (Prompt) | Qwen | 开源 | 35.5 | 86.4 | 74.7 | 20.5 | 4.5 | 23.9 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 56 | Arch-Agent-3B | katanemo | 开源 | 35.4 | 86.7 | 72.9 | 34.9 | 0.5 | 6.9 | 官方榜单 | 2026-10-22 | 许可证:katanemo-research |
| 57 | Claude-Opus-4-5-20251101 (Prompt) | Anthropic | 闭源 | 33.5 | 89.7 | 76.0 | 16.1 | 13.0 | 1.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 58 | GPT-4.1-nano-2025-04-14 (FC) | OpenAI | 闭源 | 33.0 | 73.0 | 60.8 | 23.6 | 11.0 | 18.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 59 | Mistral-Small-2506 (Prompt) | Mistral AI | 闭源 | 32.4 | 89.7 | 79.0 | 14.8 | 7.5 | 15.1 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 60 | Arch-Agent-1.5B | katanemo | 开源 | 32.1 | 82.7 | 67.7 | 26.6 | 0.0 | 8.2 | 官方榜单 | 2026-10-22 | 许可证:katanemo-research |
| 61 | Command R7B (FC) | Cohere | 开源 | 32.1 | 81.0 | 69.1 | 8.3 | 27.0 | 5.2 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 62 | Llama-3.3-70B-Instruct (FC) | Meta | 开源 | 31.9 | 88.0 | 76.6 | 21.5 | 10.0 | 8.2 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 3 Community |
| 63 | mistral-large-2411 (Prompt) | Mistral AI | 闭源 | 31.8 | 83.0 | 68.1 | 13.8 | 20.0 | 23.7 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 64 | Hammer2.1-7b (FC) | MadeAgents | 开源 | 31.7 | 85.5 | 69.5 | 23.9 | 0.0 | 0.0 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 65 | xLAM-2-1b-fc-r (FC) | Salesforce | 开源 | 30.4 | 69.0 | 55.1 | 36.0 | 0.0 | 3.9 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 66 | Gemma-3-12b-it (Prompt) | 开源 | 30.4 | 79.4 | 74.2 | 5.8 | 4.0 | 27.5 | 官方榜单 | 2026-10-22 | 许可证:gemma-terms-of-use | |
| 67 | GPT-4.1-mini-2025-04-14 (Prompt) | OpenAI | 闭源 | 29.7 | 84.6 | 74.8 | 2.5 | 4.0 | 24.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 68 | Hammer2.1-3b (FC) | MadeAgents | 开源 | 29.7 | 85.0 | 70.5 | 16.5 | 0.0 | 3.0 | 官方榜单 | 2026-10-22 | 许可证:qwen-research |
| 69 | Gemma-3-27b-it (Prompt) | 开源 | 29.5 | 87.2 | 74.5 | 10.8 | 0.0 | 13.6 | 官方榜单 | 2026-10-22 | 许可证:gemma-terms-of-use | |
| 70 | Phi-4 (Prompt) | Microsoft | 开源 | 28.8 | 69.6 | 60.7 | 3.9 | 4.5 | 24.7 | 官方榜单 | 2026-10-22 | 许可证:MIT |
| 71 | Qwen3-1.7B (FC) | Qwen | 开源 | 28.4 | 82.9 | 74.6 | 11.0 | 2.5 | 6.0 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 72 | Llama-4-Scout-17B-16E-Instruct (FC) | Meta | 开源 | 28.1 | 89.4 | 74.7 | 9.0 | 14.5 | 8.2 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 4 Community |
| 73 | Gemini-2.5-Flash-Lite (Prompt) | 闭源 | 28.0 | 83.9 | 54.9 | 7.6 | 0.0 | 12.7 | 官方榜单 | 2026-10-22 | 许可证:Proprietary | |
| 74 | CoALM-70B | UIUC + Oumi | 开源 | 28.0 | 83.4 | 67.3 | 10.6 | 0.0 | 5.8 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 3 Community |
| 75 | Hammer2.1-1.5b (FC) | MadeAgents | 开源 | 27.9 | 83.0 | 69.5 | 15.6 | 0.0 | 0.0 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 76 | palmyra-x-004 (FC) | Writer | 闭源 | 27.9 | 87.5 | 77.9 | 0.4 | 2.5 | 13.1 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 77 | GPT-5-mini-2025-08-07 (Prompt) | OpenAI | 闭源 | 27.8 | 68.0 | 62.5 | 5.5 | 8.5 | 29.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 78 | Open-Mistral-Nemo-2407 (FC) | Mistral AI | 闭源 | 27.6 | 82.8 | 73.8 | 7.8 | 7.0 | 10.3 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 79 | GPT-5-nano-2025-08-07 (Prompt) | OpenAI | 闭源 | 27.6 | 80.8 | 70.7 | 0.8 | 13.5 | 24.5 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 80 | Amazon-Nova-2-Lite-v1:0 (FC) | Amazon | 闭源 | 27.1 | 87.0 | 80.8 | 2.1 | 5.0 | 2.4 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 80 | Granite-3.1-8B-Instruct (FC) | IBM | 开源 | 27.1 | 78.3 | 60.3 | 7.5 | 0.5 | 14.4 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 82 | Falcon3-10B-Instruct (FC) | TII UAE | 开源 | 27.0 | 85.0 | 75.4 | 6.5 | 1.5 | 27.5 | 官方榜单 | 2026-10-22 | 许可证:falcon-llm-license |
| 83 | Granite-3.2-8B-Instruct (FC) | IBM | 开源 | 26.9 | 79.8 | 60.3 | 7.4 | 0.5 | 12.5 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 84 | CoALM-8B | UIUC + Oumi | 开源 | 26.8 | 84.9 | 66.8 | 8.0 | 0.0 | 2.8 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 3 Community |
| 85 | Llama-3.1-8B-Instruct (Prompt) | Meta | 开源 | 25.8 | 84.0 | 70.8 | 11.1 | 3.0 | 10.8 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 3 Community |
| 86 | MiniCPM3-4B-FC (FC) | openbmb | 开源 | 25.6 | 81.8 | 65.2 | 3.9 | 0.0 | 12.0 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 87 | Claude-Haiku-4-5-20251001 (Prompt) | Anthropic | 闭源 | 25.3 | 55.4 | 52.5 | 1.8 | 19.5 | 2.6 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 88 | Amazon-Nova-Pro-v1:0 (FC) | Amazon | 闭源 | 25.0 | 86.6 | 78.5 | 1.9 | 2.5 | 1.9 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 89 | Claude-Sonnet-4-5-20250929 (Prompt) | Anthropic | 闭源 | 24.9 | 59.8 | 46.6 | 1.6 | 16.0 | 5.4 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 90 | GPT-4.1-nano-2025-04-14 (Prompt) | OpenAI | 闭源 | 24.9 | 72.4 | 50.3 | 2.0 | 1.5 | 16.8 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 91 | Falcon3-7B-Instruct (FC) | TII UAE | 开源 | 24.0 | 82.7 | 68.3 | 5.0 | 0.5 | 20.6 | 官方榜单 | 2026-10-22 | 许可证:falcon-llm-license |
| 92 | Qwen3-0.6B (FC) | Qwen | 开源 | 23.9 | 71.8 | 56.6 | 3.6 | 1.0 | 8.6 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 93 | Granite-20b-FunctionCalling (FC) | IBM | 开源 | 23.2 | 82.3 | 58.7 | 5.4 | 0.0 | 0.0 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 94 | Qwen3-0.6B (Prompt) | Qwen | 开源 | 22.4 | 70.0 | 49.4 | 1.4 | 0.5 | 8.4 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 95 | Amazon-Nova-Micro-v1:0 (FC) | Amazon | 闭源 | 22.3 | 74.1 | 66.3 | 1.4 | 1.5 | 2.4 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 96 | RZN-T (Prompt) | Phronetic AI | 开源 | 22.3 | 67.9 | 49.7 | 2.9 | 0.0 | 6.9 | 官方榜单 | 2026-10-22 | 许可证:apache-2.0 |
| 97 | MiniCPM3-4B (Prompt) | openbmb | 开源 | 22.1 | 70.5 | 43.1 | 3.5 | 2.0 | 9.5 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 98 | Llama-3.2-3B-Instruct (FC) | Meta | 开源 | 21.9 | 82.7 | 58.3 | 4.0 | 1.0 | 6.2 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 3 Community |
| 99 | Bielik-11B-v2.3-Instruct (Prompt) | SpeakLeash & ACK Cyfronet AGH | 开源 | 21.9 | 81.5 | 67.8 | 2.6 | 1.5 | 11.4 | 官方榜单 | 2026-10-22 | 许可证:Apache 2.0 |
| 100 | Hammer2.1-0.5b (FC) | MadeAgents | 开源 | 21.2 | 66.0 | 54.6 | 2.9 | 0.0 | 1.1 | 官方榜单 | 2026-10-22 | 许可证:cc-by-nc-4.0 |
| 101 | Gemma-3-4b-it (Prompt) | 开源 | 19.6 | 61.1 | 60.8 | 0.4 | 1.0 | 8.6 | 官方榜单 | 2026-10-22 | 许可证:gemma-terms-of-use | |
| 102 | Open-Mistral-Nemo-2407 (Prompt) | Mistral AI | 闭源 | 19.3 | 88.5 | 74.0 | 0.8 | 2.5 | 8.6 | 官方榜单 | 2026-10-22 | 许可证:Proprietary |
| 103 | Granite-4.0-350m (FC) | IBM | 开源 | 19.0 | 67.9 | 46.1 | 2.5 | 0.5 | 3.2 | 官方榜单 | 2026-10-22 | 许可证:Apache-2.0 |
| 104 | Falcon3-3B-Instruct (FC) | TII UAE | 开源 | 16.3 | 54.6 | 54.5 | 1.0 | 1.0 | 7.7 | 官方榜单 | 2026-10-22 | 许可证:falcon-llm-license |
| 105 | Ministral-8B-Instruct-2410 (FC) | Mistral AI | 开源 | 11.1 | 0.0 | 0.0 | 0.0 | 1.0 | 4.5 | 官方榜单 | 2026-10-22 | 许可证:Mistral AI Research License |
| 106 | Falcon3-1B-Instruct (FC) | TII UAE | 开源 | 11.1 | 9.0 | 2.9 | 0.0 | 0.0 | 5.8 | 官方榜单 | 2026-10-22 | 许可证:falcon-llm-license |
| 107 | Llama-3.2-1B-Instruct (FC) | Meta | 开源 | 10.8 | 38.4 | 11.8 | 0.0 | 0.0 | 3.2 | 官方榜单 | 2026-10-22 | 许可证:Meta Llama 3 Community |
| 108 | Llama-3.1-Nemotron-Ultra-253B-v1 (FC) | NVIDIA | 开源 | 10.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 官方榜单 | 2026-10-22 | 许可证:nvidia-open-model-license |
| 109 | Gemma-3-1b-it (Prompt) | 开源 | 7.2 | 20.2 | 11.8 | 0.0 | 0.0 | 3.2 | 官方榜单 | 2026-10-22 | 许可证:gemma-terms-of-use | |
榜单说明
第 4 批接入Berkeley Function Calling Leaderboard(BFCL)由 UC Berkeley Gorilla 团队维护,评测大模型在可执行函数调用场景中的正确率。当前公开版本为 V4,在 v1–v3 的单轮/多轮调用之外,加入 Web Search、Memory 等 Agentic 子类,以及(对 Prompt 路线)格式敏感度相关设置。
榜单页面说明:总分 Overall Accuracy 为各子类分数的未加权平均;模型名中的 FC 表示原生 function/tool calling,Prompt 表示用普通文本生成模拟调用。页面标注最近更新为 2026-04-12,复现建议使用仓库 commit f7cf735 或 PyPI 包 bfcl-eval==2025.12.17。
本站每日同步官方页面加载的 data_overall.csv(与页面表格同源),收录全部模型;子项列为 CSV 中的 Non-Live、Live、Multi-Turn、Web Search、Memory 汇总准确率。许可证为 Proprietary 的记为闭源权重,其余具名许可证(含研究或非商用许可)记为开源权重。不同 FC/Prompt 设置为不同条目,不在此合并。评测代码与数据在 Gorilla 仓库的 berkeley-function-call-leaderboard 目录。
- 指标
- Overall Accuracy:各评测子类准确率的未加权平均(%)(越高越好)
- 子项
- 单轮Non-LiveLive多轮Multi-TurnAgenticWeb SearchMemory
- 数据源
- BFCL 官方榜单(V4)(在新窗口打开)(官方榜单)
- 来源类型
- 官方榜单 109
分数取自数据源原文;同一模型在不同评测设置或不同来源下分别列出,来源冲突时全部保留并标注来源类型。
更新记录
0 条首次同步作为基线,之后新上榜、分数变化和移除的条目会记录在这里。