Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use
Specification
most-agreed values
Context
164K
Max output
82K
Released
2025-12-01
Knowledge cutoff
2024-07
Retires
—
Open weights
yes
Input
text
Output
text
Price
US dollars per million tokens · most-agreed
Input
$0.62
Output
$1.85
Available from 5 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| amazon-bedrock | $0.62 | $1.85 | — | — | 164K | 82K | — |
| bedrock_converse | $0.62 | $1.85 | — | — | 164K | 164K | — |
| bedrock_converse · us | $0.62 | $1.85 | — | — | 164K | 164K | — |
| bedrock | $0.74 $0.62 | $2.22 $1.85 | — | — | 164K | 164K | — |
| bedrock_converse · eu | $0.74 | $2.22 | — | — | 164K | 164K | — |
Other listings of this model
this page is one of them
The corpus files DeepSeek V3.2 under several keys. This page is the v3-2 listing. The full record — every host, every price and every benchmark score — is on the main DeepSeek V3.2 page.
More from DeepSeek
most-hosted first