Model Detail
DeepSeek V3.2 (thinking)
DeepSeek · Text · チャット補完
Lightweight model for cost-sensitive chat, basic generation, and batch text tasks.
Updated 2026-03-21
Input
0.28 USD / 1M tokens
Output
0.42 USD / 1M tokens
Context window
128,000 tokens
Max output
Not provided
Capabilities
Knowledge cutoff
Not provided
Reasoning model
Yes
Primary unit
1M トークン
Official and future third-party prices are shown in a shared unit.
Evidence rows
1
Each price row keeps source site, observed time, and excerpt.
Quick verify
Verify on Artificial Analysis
Start from this summary, then cross-check official sources or leaderboards.
Use cases
Good for
Not ideal for
Artificial Analysis snapshot
Fields below come from the AA model page for quick cross-checking.
AA Intelligence
41.71
AA Input price
3 USD / 1M tokens
AA Output price
4.5 USD / 1M tokens
Input modalities
Text
Output modalities
Text
Source domain
artificialanalysis.aiPricing details
Structured official and third-party prices with extensible dimensions.
Official pricing
Third-party provider pricing
Sources and evidence
Each price line includes a source chain for trust and verification.
Field入力 / 出力
Scenarioチャット補完
Captured value0.28 USD / 1M tokens
Source domainapi-docs.deepseek.com
Observed at2026-03-21
Excerpt1M INPUT TOKENS (CACHE MISS) — deepseek-reasoner (per 1M tokens, vendor table) --- 1M INPUT TOKENS (CACHE HIT) — deepseek-reasoner (per 1M tokens, vendor table) --- 1M OUTPUT TOKENS — deepseek-reasoner (per 1M tokens, vendor table)