DeepSeek V3.2
DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...
The figures
- Made by
- DeepSeek
- Model ID
- deepseek/deepseek-v3.2
- Input
- $0.28 per million tokens
- Output
- $0.42 per million tokens
- Context window
- 164K tokens
- Longest answer
- 66K tokens
- Takes in
- text
- Gives back
- text
- Reasoning
- Yes
- Tool calling
- Yes
- Reads images
- No
- Open weights
- Yes
- Released
- 1 Dec 2025
Read on 3 Oct 2026.