LLM Leaderboard
Last updated: 48m ago
这是面向 “llm leaderboard”“ai ranking”“ai rankings” 等核心查询的专用着陆页，强调当前榜单顺序、头部模型差距与近期真实变动。
前 20 名 AI 模型综合评分
基于基准测试、能力和上下文窗口排名
LMMarketCap.com
当前前 25 名
| # | 模型 | 服务商 | 评分 | 7天变动 | 输入价格 |
|---|---|---|---|---|---|
| #1 | Claude Fable 5 | Anthropic | 97 | 0 | $10.00/M |
| #2 | Claude Fable 5 (batch) | Anthropic | 97 | 0 | $5.00/M |
| #3 | Claude Opus 5 (Fast) | Anthropic | 95 | +228 | $10.00/M |
| #4 | Claude Opus 5 | Anthropic | 95 | +228 | $5.00/M |
| #5 | Claude Opus 4.8 (Fast) | Anthropic | 95 | +1 | $10.00/M |
| #6 | Claude Opus 4.8 | Anthropic | 95 | +1 | $5.00/M |
| #7 | Claude Opus 4.7 (Fast) | Anthropic | 95 | -4 | $30.00/M |
| #8 | Claude Opus 4.7 | Anthropic | 95 | -4 | $5.00/M |
| #9 | Claude Opus 4.7 (batch) | Anthropic | 95 | -4 | $2.50/M |
| #10 | Claude Opus 4.8 (batch) | Anthropic | 95 | -2 | $2.50/M |
| #11 | GPT-5.5 Pro | OpenAI | 93 | +8 | $30.00/M |
| #12 | GPT-5.5 Pro (batch) | OpenAI | 93 | -3 | $15.00/M |
| #13 | GPT-5.5 | OpenAI | 93 | -3 | $5.00/M |
| #14 | GPT-5.5 (batch) | OpenAI | 93 | -3 | $2.50/M |
| #15 | Gemini 3.1 Pro Preview Custom Tools | 92 | -3 | $2.00/M | |
| #16 | Gemini 3.1 Pro Preview | 92 | -3 | $2.00/M | |
| #17 | Gemini 3.1 Pro Preview (batch) | 92 | -3 | $1.00/M | |
| #18 | GPT-5.4 Pro | OpenAI | 92 | -3 | $30.00/M |
| #19 | GPT-5.4 Pro (batch) | OpenAI | 92 | -3 | $15.00/M |
| #20 | GPT-5.4 | OpenAI | 92 | -3 | $2.50/M |
| #21 | GPT-5.4 (batch) | OpenAI | 92 | -3 | $1.25/M |
| #22 | GPT-5.3-Codex | OpenAI | 91 | +17 | $1.75/M |
| #23 | GPT-5.2-Codex | OpenAI | 91 | +169 | $1.75/M |
| #24 | GPT-5.2 Chat | OpenAI | 91 | +77 | $1.75/M |
| #25 | GPT-5.2 Pro | OpenAI | 91 | -5 | $21.00/M |
最大上升者
配套页面
Frequently Asked Questions
This page ranks live coding and language models using LM Market Cap composite scores, which combine capabilities, pricing, context, output capacity, versatility, and recency into a single leaderboard surface.
The models directory is for browsing the full catalog across providers and categories. This page is the competitive LLM ranking page aimed at users who want the top-performing models in rank order.
Yes. The linked rank-change and trends surfaces now compare live rankings against the latest archived weekly snapshot rather than synthetic movement estimates.