DeepSeek-V3.2 at a glance
DeepSeek-V3.2 became the default model behind its app and API, with thinking built into tool use, while V3.2-Speciale was offered briefly for maximum reasoning.
What changed with DeepSeek-V3.2
On December 1, 2025, DeepSeek released DeepSeek-V3.2 as the successor to V3.2-Exp and upgraded its deepseek-chat and deepseek-reasoner API names to it. Both V3.2 and V3.2-Speciale are open-sourced on Hugging Face.
DeepSeek calls V3.2 the first model to fold thinking directly into tool use, trained on agent data covering more than 1,800 environments. V3.2-Speciale targets maximum reasoning, does not support tool calls, and ran on a temporary API endpoint until December 15, 2025.
Where to compare DeepSeek-V3.2
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
Leaderboards
General leaderboards
Why DeepSeek-V3.2 matters for businesses
For automation work, V3.2 offered a low-cost, open-weight reasoning model with agent-oriented training, though businesses should weigh data-residency requirements before sending client data to any hosted API.
Sources for this story
- DeepSeek-V3.2 release (opens in a new tab)Official
DeepSeek API Docs · api-docs.deepseek.com
- Change Log (opens in a new tab)Official
DeepSeek API Docs · api-docs.deepseek.com
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
- #reasoning