OpenRouter 目前提供 22 个大模型 API。
速度基准测试平均吞吐 85 tok/s。
https://openrouter.ai| 时间 | 模型 | 速度 | 延迟 |
|---|---|---|---|
| Apr 13, 03:04 AM | z-ai/glm-5.1 | 63.01 tok/s | 12.13s |
| Apr 12, 11:56 AM | openai/gpt-oss-120b:free | 34.72 tok/s | 1.75s |
| Apr 11, 04:25 AM | arcee-ai/trinity-large-preview:free | 5.85 tok/s | 1.51s |
| Apr 11, 04:15 AM | arcee-ai/trinity-large-preview:free | 5.83 tok/s | 1.18s |
| Apr 8, 02:33 PM | z-ai/glm-5.1 | 43.91 tok/s | 14.92s |
| Apr 8, 02:30 PM | z-ai/glm-5.1 | 67.31 tok/s | 15.10s |
| Apr 7, 06:24 AM | nvidia/nemotron-3-super-120b-a12b:free | 22.52 tok/s | 33.28s |
| Apr 7, 06:22 AM | stepfun/step-3.5-flash:free | 78.32 tok/s | 6.69s |
| Apr 3, 12:56 PM | qwen/qwen3.6-plus:free | 47.59 tok/s | 24.10s |
| Apr 3, 12:45 PM | qwen/qwen3.6-plus:free | 46.82 tok/s | 26.63s |