选择一个模型后会直接打开对应的对比页面。
DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workfl...
Input and output token limits for this model, plus how it ranks on long-context understanding.
来自 OpenRouter 的第三方端点数据，和 LMSpeed 自测数据分开展示。部分 30 分钟实时性能字段需要配置 OpenRouter API key 后同步才会出现。
| Provider endpoint | 输入 | 输出 | 1 天在线率 | 30m 延迟 | 30m 吞吐 | 上下文 / 输出 |
|---|---|---|---|---|---|---|
DeepSeek deepseek/fp8 | $0.140/M | $0.280/M | 100.0% | — | — | 1.0M tokens / 384K tokens |
GMICloud gmicloud/fp8 | $0.140/M | $0.280/M | 100.0% | — | — | 1.0M tokens / — |
Cloudflare cloudflare/fp8 | $0.140/M | $0.280/M | 99.9% | — | — | 384K tokens / 384K tokens |
SiliconFlow siliconflow/fp8 | $0.140/M | $0.280/M | 99.8% | — | — | 1.0M tokens / 393.2K tokens |
DeepInfra deepinfra/fp4 | $0.090/M | $0.180/M | 95.3% | — | — | 1.0M tokens / 65.5K tokens |
Parasail parasail/fp8 | $0.140/M | $0.280/M | 92.5% | — | — | 1.0M tokens / 1.0M tokens |
AtlasCloud atlas-cloud/fp8 | $0.140/M | $0.280/M | 84.2% | — | — | 262.1K tokens / 131.1K tokens |
Fireworks fireworks | $0.140/M | $0.280/M | 66.1% | — | — | 1.0M tokens / — |
