DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...
| Field | Per token (USD) | Per million |
|---|---|---|
| prompt | 0.00000025 | $0.250 |
| completion | 0.00000095 | $0.950 |
| input_cache_read | 0.00000013 | $0.130 |