DeepSeek-V3.2-Exp at a glance
DeepSeek-V3.2-Exp introduced DeepSeek Sparse Attention for cheaper long-context work, matched V3.1-Terminus on benchmarks, and came with an API price cut of 50% or more.
What changed with DeepSeek-V3.2-Exp
On September 29, 2025, DeepSeek released DeepSeek-V3.2-Exp, an experimental model built on V3.1-Terminus. Its key change is DeepSeek Sparse Attention, a fine-grained sparse attention method meant to make training and inference on long contexts faster and cheaper with minimal impact on output quality.
DeepSeek said benchmark results were on par with V3.1-Terminus, made the model live in its app, web and API, and cut API prices by more than 50% effective immediately. Weights, a technical report and GPU kernels were open-sourced, and V3.1-Terminus stayed on a temporary endpoint until October 15, 2025 for comparison.
Where to compare DeepSeek-V3.2-Exp
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
General leaderboards
Why DeepSeek-V3.2-Exp matters for businesses
For businesses, the price cut showed that architecture changes, not only newer models, keep pushing AI costs down, which makes long-document processing and high-volume automations cheaper to run.
Sources for this story
- Introducing DeepSeek-V3.2-Exp (opens in a new tab)Official
DeepSeek API Docs · api-docs.deepseek.com
- deepseek-ai/DeepSeek-V3.2-Exp (opens in a new tab)Official
Hugging Face · huggingface.co
- DeepSeek V3.2 Exp model page (opens in a new tab)Benchmark
Artificial Analysis · artificialanalysis.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #deepseek
- #open-weights
- #long-context