模型排行
769 个模型 · 5 个榜单 · 2 个数据来源
智力前三甲
GPT-5.2 (xhigh)OpenAI50.6
Claude Opus 4.5 (Reasoning)Anthropic49.1
Gemini 3 Pro Preview (high)Google47.9
开源前三甲
GLM-4.7 (Reasoning)Z AI · 总第 9 名41.7
DeepSeek V3.2 (Reasoning)DeepSeek · 总第 12 名41.2
Kimi K2 ThinkingKimi · 总第 14 名40.3
这一期的变化
gpt-5.2-codex编程榜 · 第 17新上榜
FLUX.2 [klein] 9B图像榜 · 第 21新上榜
FLUX.2 [klein] Base 9B图像榜 · 第 43新上榜
FLUX.2 [klein] 4B图像榜 · 第 48新上榜
GLM-Image图像榜 · 第 49新上榜
| 排名 | 变化 | 模型 | 智力 | 价格 $/M | 速度 tok/s | 首字延迟 |
|---|---|---|---|---|---|---|
| 1 | — | GLM-4.7 (Reasoning)开源Z AI · 另有 1 个档位 | 41.7 | — | — | — |
| 2 | — | DeepSeek V3.2 (Reasoning)开源DeepSeek · 另有 1 个档位 | 41.2 | — | — | — |
| 3 | — | Kimi K2 Thinking开源Kimi | 40.3 | — | — | — |
| 4 | — | MiniMax-M2.1开源MiniMax | 39.4 | — | — | — |
| 5 | — | MiMo-V2-Flash (Reasoning)开源Xiaomi · 另有 1 个档位 | 39.1 | — | — | — |
| 6 | — | MiniMax-M2开源MiniMax | 35.7 | — | — | — |
| 7 | — | DeepSeek V3.1 Terminus (Reasoning)开源DeepSeek · 另有 1 个档位 | 33.5 | — | — | — |
| 8 | — | gpt-oss-120B (high)开源OpenAI · 另有 1 个档位 | 32.9 | — | — | — |
| 9 | — | DeepSeek V3.2 Exp (Reasoning)开源DeepSeek · 另有 1 个档位 | 32.5 | — | — | — |
| 10 | — | GLM-4.6 (Reasoning)开源Z AI · 另有 1 个档位 | 32.2 | — | — | — |
| 11 | — | K-EXAONE (Reasoning)开源LG AI Research · 另有 1 个档位 | 31.9 | — | — | — |
| 12 | — | Qwen3 235B A22B 2507 (Reasoning)开源Alibaba · 另有 2 个档位 | 29.4 | — | — | — |
| 13 | — | DeepSeek V3.2 Speciale开源DeepSeek | 29.0 | — | — | — |
| 14 | — | Kimi K2 0905开源Kimi | 28.0 | — | — | — |
| 15 | — | DeepSeek V3.1 (Reasoning)开源DeepSeek · 另有 1 个档位 | 27.9 | — | — | — |
| 16 | — | Apriel-v1.6-15B-Thinker开源ServiceNow | 27.8 | — | — | — |
| 17 | — | Qwen3 VL 235B A22B (Reasoning)开源Alibaba | 27.5 | — | — | — |
| 18 | — | DeepSeek R1 0528 (May '25)开源DeepSeek · 另有 1 个档位 | 27.1 | — | — | — |
| 19 | — | GLM-4.5 (Reasoning)开源Z AI | 26.8 | — | — | — |
| 20 | — | Qwen3 Next 80B A3B (Reasoning)开源Alibaba | 26.5 | — | — | — |
| 21 | — | Seed-OSS-36B-Instruct开源ByteDance Seed | 25.0 | — | — | — |
| 22 | — | Qwen3 Coder 480B A35B Instruct开源Alibaba | 24.8 | — | — | — |
| 23 | — | NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)开源NVIDIA · 另有 1 个档位 | 24.7 | — | — | — |
| 24 | — | gpt-oss-20B (high)开源OpenAI · 另有 1 个档位 | 24.7 | — | — | — |
| 25 | — | Qwen3 235B A22B 2507 Instruct开源Alibaba | 24.6 | — | — | — |
| 26 | — | MiniMax M1 80k开源MiniMax | 24.3 | — | — | — |
| 27 | — | HyperCLOVA X SEED Think (32B)开源Naver | 24.0 | — | — | — |
| 28 | — | GLM-4.6V (Reasoning)开源Z AI · 另有 1 个档位 | 23.8 | — | — | — |
| 29 | — | GLM-4.5-Air开源Z AI | 23.6 | — | — | — |
| 30 | — | Mi:dm K 2.5 Pro开源Korea Telecom | 23.2 | — | — | — |
| 31 | — | Qwen3 30B A3B 2507 (Reasoning)开源Alibaba · 另有 2 个档位 | 22.8 | — | — | — |
| 32 | — | DeepSeek V3 0324开源DeepSeek · 另有 1 个档位 | 22.6 | — | — | — |
| 33 | — | Mistral Large 3开源Mistral | 22.5 | — | — | — |
| 34 | — | INTELLECT-3开源Prime Intellect | 22.2 | — | — | — |
| 35 | — | Devstral 2开源Mistral | 21.7 | — | — | — |
| 36 | — | K2-V2 (high)开源MBZUAI Institute of Foundation Models · 另有 2 个档位 | 21.1 | — | — | — |
| 37 | — | Qwen3 VL 235B A22B Instruct开源Alibaba | 20.8 | — | — | — |
| 38 | — | Qwen3 Next 80B A3B Instruct开源Alibaba | 20.3 | — | — | — |
| 39 | — | Qwen3 Coder 30B A3B Instruct开源Alibaba | 20.2 | — | — | — |
| 40 | — | Ling-1T开源InclusionAI | 19.7 | — | — | — |
| 40 | — | Qwen3 VL 30B A3B (Reasoning)开源Alibaba | 19.7 | — | — | — |
| 42 | — | Llama Nemotron Super 49B v1.5 (Reasoning)开源NVIDIA · 另有 1 个档位 | 19.4 | — | — | — |
| 43 | — | Devstral Small 2开源Mistral | 19.1 | — | — | — |
| 44 | — | Llama 4 Maverick开源Meta | 18.9 | — | — | — |
| 45 | — | Qwen3 4B 2507 (Reasoning)开源Alibaba | 18.8 | — | — | — |
| 46 | — | Llama 3.1 Instruct 405B开源Meta | 17.5 | — | — | — |
| 47 | — | Qwen3 VL 32B Instruct开源Alibaba | 17.1 | — | — | — |
| 48 | — | Qwen3 32B (Reasoning)开源Alibaba | 17.1 | — | — | — |
| 49 | — | Qwen3 VL 8B (Reasoning)开源Alibaba | 17.0 | — | — | — |
| 50 | — | EXAONE 4.0 32B (Reasoning)开源LG AI Research · 另有 1 个档位 | 17.0 | — | — | — |
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