DeepSeek-V4-Pro (Preview)

Open weights Active Text LLM MoE deepseek-v4-pro

Preview of the V4 flagship: 1.6T-total/49B-active MoE with 1M context and hybrid sparse attention.

Released
Apr 24, 2026
Context window
1M tokens
Max output
Not disclosed
Parameters
1,600B 49B active
Input price
Not disclosed
Output price
Not disclosed

Prices are first-party list prices in USD at the standard tier. Last verified Oct 30, 2026.

Pricing

Launch priceApr 24, 2026
$1.74 in / $3.48 out
Current price
No longer sold

Specifications

Developer
DeepSeek
Model type
Text LLM
Architecture
MoE
Parameters
1,600B (49B active)
Input
Text
Output
Text
Knowledge cutoff
Not disclosed
License
MIT License
API access
No first-party API
Capabilities
Agentic Coding Flagship Long context Reasoning