DeepSeek V3.1

Text Generation
DeepSeek · deepseek-v3.1

840B param hybrid with dual-mode (thinking/non-thinking) - cached input $0.07/M

Context Window

128,000

tokens

Max Output

8,192

tokens

Input Pricing

$0.56

per million tokens

Output Pricing

$1.68

per million tokens

Core Capabilities

Vision Support
Function Calling
Streaming

Additional Capabilities

Cached Input

Cache hits at $0.07/M - 87.5% discount

Dual Mode

Hybrid model with both thinking and non-thinking modes

No benchmark scores available

Released: August 01, 2025