YZ Index
AI 价格追踪器
主要AI模型的API价格 × 评测实际成本。降价·涨价一目了然
| # | 模型 | 服务商 | 入力价格 $/1M tokens | 出力价格 $/1M tokens | 评测实际成本 平均/题 | 评测总成本 |
|---|---|---|---|---|---|---|
| 1 | 豆包 Pro | doubao | $0.80 | $2.00 | $0.0014 | $0.140 (100問) |
| 2 | Gemini 2.5 Pro | gemini | $1.25 | $10.00 | $0.0056 | $0.562 (100問) |
| 3 | Qwen3 Max | qwen | $1.60 | $6.40 | $0.0040 | $0.400 (100問) |
| 4 | DeepSeek V4 Pro | deepseek | $2.00 | $8.00 | $0.0050 | $0.500 (100問) |
| 5 | GLM-4.6 | zhipu | $2.00 | $8.00 | $0.0050 | $0.475 (95問) |
| 6 | Gemini 3.1 Pro | gemini | $2.50 | $15.00 | $0.0088 | $0.875 (100問) |
| 7 | Claude Sonnet 4.6 | claude | $3.00 | $15.00 | $0.0090 | $0.900 (100問) |
| 8 | Grok 4 | grok | $3.00 | $15.00 | $0.0090 | $0.900 (100問) |
| 9 | GPT-5.5 | gpt | $5.00 | $20.00 | $0.0125 | $1.250 (100問) |
| 10 | GPT-o3 | gpt | $10.00 | $40.00 | $0.0250 | $2.500 (100問) |
| 11 | Claude Opus 4.7 | claude | $15.00 | $75.00 | $0.0450 | $4.500 (100問) |
价格来自各服务商官方API价格页面。评测实际成本由最新完整评测（Run #259）中各题目的 cost_est 字段计算得出。
价格变动箭头由 yz_pricing_history 表自动检测。评测执行时会记录快照。