DeepSeek-V3.2-Exp
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
DeepSeek-V3.2-Exp — the first to use DeepSeek Sparse Attention
Why it matters
Puts fine-grained sparse attention into a commercial API and cuts long-context costs
Source — Primary sources first; secondary reports are labelled “Press”
Officialapi-docs.deepseek.comhttps://api-docs.deepseek.com/news/news250929Added in Week 35, 2026 · See that week
On the map — ⑧ Inference Infrastructure line
- Previous stopGemma 3 QAT int4
- This record · stopDeepSeek-V3.2-Exp
- Next stopIronwood
In the model family tree — DeepSeek family · 12 versions
- Earlier versionDeepSeek-V3.1
- This recordDeepSeek-V3.2-Exp
- Later versionDeepSeek-V3.2