DeepSeek: DeepSeek V3.1
deepseek/deepseek-chat-v3.1
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...
Modalities
In / Out Price
$0.25 / $0.95 per 1M
Context
164K
Released
Aug 21, 2025
Knowledge Cutoff
Mar 2025
Providers
Different carriers host the same model. Reroute routes each request to the best one on price and uptime, and fails over to the next if a carrier goes down.
| Provider | Context | Input /M | Output /M | Cache read /M | Latency | Throughput | Uptime |
|---|---|---|---|---|---|---|---|
 DeepInfrafp4 | 164K | $0.25 | $0.95 | $0.13 | — | — | 100.00% |
 SiliconFlowfp8 | 164K | $0.27 | $1 | — | — | — | 100.00% |
 CoreWeavefp8 | 161K | $0.55 | $1.65 | $0.55 | — | — | 100.00% |
 Mara | 131K | $0.60 | $1.70 | — | — | — | 96.55% |
 SambaNovafp8 | 131K | $0.65 | $1.50 | — | — | — | 100.00% |