Skip to main content
DeepSeek exposes its current Flash and Pro models through an OpenAI-compatible API. Roomote connects to that API directly rather than routing requests through another inference provider.

​
Get an API key

Create a key in the DeepSeek platform and add balance to the account that will run Roomote tasks.

​
Configuration

Add DeepSeek in Settings > Models and paste the key, or configure:
Roomote recommends these direct model routes:
  • deepseek/deepseek-v4-pro for coding, code review, and planning
  • deepseek/deepseek-flash for helper, explore, and vision work
The Flash model accepts text and image input. The Pro model accepts text input. Both models support tool calls and configurable reasoning.

​
Cost behavior

DeepSeek charges the connected account for input, cached input, and output tokens. Roomote records usage and estimates cost from current model metadata; DeepSeek’s models and pricing page remains authoritative.

​
Verify setup

  1. save DEEPSEEK_API_KEY
  2. confirm DeepSeek V4.1 Flash and DeepSeek V4 Pro 0813 appear in Settings > Models
  3. confirm the desired models and role mapping are enabled
  4. run a small task and confirm the request appears in DeepSeek platform usage

​
Common issues

  • Authentication or balance errors occur. Confirm the key is active and the account has enough balance.
  • A model ID is rejected. Use the exact deepseek/... ID shown in Roomote; OpenRouter and other provider routes are separate connections.
  • Image input fails on Pro. Assign deepseek/deepseek-flash to the vision role; the Pro model is text-only.