AI
best practice
Use 'Non-Thinking' Mode for Faster Response Times
Choose deepseek-chat for high-speed, everyday productivity tasks.
Switch to the `deepseek-chat` model for everyday tasks like drafting emails, summarizing text, or basic translations to minimize latency.
DeepSeek
The Scenario
You need to draft a quick email, summarize a standard meeting transcript, or perform a simple translation where you prioritize speed over deep reasoning.
Before & after
The old way
Users often wait for a 'thinking' model to process simple requests, wasting 1–2 minutes of idle time per query.
With AI
Set your model parameter to `deepseek-chat` to get immediate answers in under 30 seconds for routine business tasks.
The Prompt
Draft a professional email responding to this meeting request using your non-thinking mode for speed: [INSERT_DETAILS]
The non-thinking mode is optimized for standard chat, text transformation, and simple questions where deep internal verification isn't required, saving on latency.
Source
Change Log | DeepSeek API Docs"`deepseek-chat` corresponds to DeepSeek-V3.2's non-thinking mode."