模型排行
315 个模型 · 2 个榜单 · 2 个数据来源
智力王座
开源之光
国产之光
开源前三甲
Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)NVIDIA · 总第 7 名60.8
DeepSeek R1DeepSeek · 总第 9 名60.2
QwQ 32BAlibaba · 总第 10 名58.1
| 排名 | 变化 | 模型 | 智力 | 价格 $/M | 速度 tok/s | 首字延迟 |
|---|---|---|---|---|---|---|
| 1 | — | Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)开源NVIDIA | 60.8 | — | — | — |
| 2 | — | DeepSeek R1开源DeepSeek | 60.2 | — | — | — |
| 3 | — | QwQ 32B开源Alibaba | 58.1 | — | — | — |
| 4 | — | DeepSeek V3 0324 (Mar' 25)开源DeepSeek · 另有 1 个档位 | 53.2 | — | — | — |
| 5 | — | DeepSeek R1 Distill Qwen 32B开源DeepSeek | 51.9 | — | — | — |
| 6 | — | Llama 4 Maverick开源Meta | 50.5 | — | — | — |
| 7 | — | DeepSeek R1 Distill Qwen 14B开源DeepSeek | 49.0 | — | — | — |
| 8 | — | DeepSeek R1 Distill Llama 70B开源DeepSeek | 48.2 | — | — | — |
| 9 | — | Reka Flash 3开源Reka AI | 47.1 | — | — | — |
| 10 | — | Llama 4 Scout开源Meta | 43.0 | — | — | — |
| 11 | — | QwQ 32B-Preview开源Alibaba | 42.6 | — | — | — |
| 12 | — | Llama 3.3 Instruct 70B开源Meta | 41.1 | — | — | — |
| 13 | — | Llama 3.1 Instruct 405B开源Meta | 40.5 | — | — | — |
| 14 | — | Qwen2.5 Instruct 72B开源Alibaba | 40.4 | — | — | — |
| 15 | — | MiniMax-Text-01开源MiniMax | 40.2 | — | — | — |
| 16 | — | Phi-4开源Microsoft Azure | 40.2 | — | — | — |
| 17 | — | Command A开源Cohere | 40.0 | — | — | — |
| 18 | — | Llama 3.1 Tulu3 405B开源Allen Institute for AI | 39.9 | — | — | — |
| 19 | — | Llama 3.3 Nemotron Super 49B v1开源NVIDIA | 39.3 | — | — | — |
| 20 | — | Mistral Large 2 (Nov '24)开源Mistral · 另有 1 个档位 | 38.3 | — | — | — |
| 21 | — | Gemma 3 27B Instruct开源Google | 37.6 | — | — | — |
| 22 | — | Pixtral Large开源Mistral | 37.4 | — | — | — |
| 23 | — | Qwen2.5 Instruct 32B开源Alibaba | 37.4 | — | — | — |
| 24 | — | Llama 3.1 Nemotron Instruct 70B开源NVIDIA | 37.3 | — | — | — |
| 25 | — | Qwen2.5 Coder Instruct 32B开源Alibaba | 36.3 | — | — | — |
| 26 | — | Llama 3.1 Instruct 70B开源Meta | 35.4 | — | — | — |
| 27 | — | Mistral Small 3.1开源Mistral | 35.3 | — | — | — |
| 28 | — | Mistral Small 3开源Mistral | 35.3 | — | — | — |
| 29 | — | DeepSeek R1 Distill Llama 8B开源DeepSeek | 34.0 | — | — | — |
| 30 | — | Gemma 3 12B Instruct开源Google | 33.8 | — | — | — |
| 31 | — | Llama 3.2 Instruct 90B (Vision)开源Meta | 33.4 | — | — | — |
| 32 | — | Qwen2 Instruct 72B开源Alibaba | 32.6 | — | — | — |
| 33 | — | DeepHermes 3 - Mistral 24B Preview开源Nous Research | 30.0 | — | — | — |
| 34 | — | Jamba 1.5 Large开源AI21 Labs | 29.3 | — | — | — |
| 35 | — | Hermes 3 - Llama-3.1 70B开源Nous Research | 29.1 | — | — | — |
| 36 | — | Jamba 1.6 Large开源AI21 Labs | 28.8 | — | — | — |
| 37 | — | Llama 3 Instruct 70B开源Meta | 27.5 | — | — | — |
| 38 | — | Mistral Small (Sep '24)开源Mistral | 27.4 | — | — | — |
| 39 | — | Phi-4 Multimodal Instruct开源Microsoft Azure | 26.9 | — | — | — |
| 40 | — | Qwen2.5 Coder Instruct 7B开源Alibaba | 26.7 | — | — | — |
| 41 | — | Mixtral 8x22B Instruct开源Mistral | 26.2 | — | — | — |
| 42 | — | Phi-4 Mini Instruct开源Microsoft Azure | 26.0 | — | — | — |
| 43 | — | Phi-3 Medium Instruct 14B开源Microsoft Azure | 24.5 | — | — | — |
| 44 | — | Gemma 3 4B Instruct开源Google | 24.5 | — | — | — |
| 45 | — | Llama 3.1 Instruct 8B开源Meta | 23.7 | — | — | — |
| 46 | — | Pixtral 12B (2409)开源Mistral | 23.4 | — | — | — |
| 47 | — | Ministral 8B开源Mistral | 22.3 | — | — | — |
| 48 | — | Gemma 2 9B开源Google | 22.2 | — | — | — |
| 49 | — | Phi-3 Mini Instruct 3.8B开源Microsoft Azure | 22.1 | — | — | — |
| 50 | — | Command-R+ (Aug '24)开源Cohere · 另有 3 个档位 | 21.5 | — | — | — |
智力进步
涨得最多 · Anthropic +57.2目前最高 · Anthropic 64.6最高与最低相差 18.1 分时间范围 2025年11月 – 2026年11月
64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi