DeepSeek V3.2
by DeepSeek
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 At
Models
DeepSeek V3.2
Added 10 July 2026
Overview
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...
Best for
Best for
High-volume cheap coding and chat
Use cases
- coding assistants
- batch analysis
- draft generation
How to use / API access
Call via OpenRouter with model id `deepseek/deepseek-v3.2`.
- openrouter ·
deepseek/deepseek-v3.2Docs
OpenRouter id: deepseek/deepseek-v3.2
Benchmarks
- Intelligence Index (artificial-analysis) : 32 index
- Coding Index (artificial-analysis) : 44.2 index
- Agentic Index (artificial-analysis) : 18.3 index
- Design Arena 3d (other) : 1209 elo · #43
- Design Arena asciiart (other) : 1130 elo · #43
- Design Arena codecategories (other) : 1207 elo · #50
- Design Arena dataviz (other) : 1200 elo · #50
Enterprise DNA angle
A/B DeepSeek V3 against Sonnet inside Omni with shared eval harnesses.
Notes
DeepSeek V3.2 is listed in the EDNA Models directory from the OpenRouter catalogue.
Pros
- Strong quality at aggressive pricing
- Useful coding and general assistant workloads
Cons
- Validate safety/brand tone before customer-facing use
- Ecosystem tooling less mature than US labs