DeepSeek
DeepSeek is the Chat engine for DeepSeek models on the Alibaba Model Studio cloud plan, in Chat — /chat?engine=deepseek.

What it does
Section titled “What it does”A dedicated engine talking to the Model Studio chat API, with the shared Chat / History / Workspace tabs and Settings panel. This is the cloud DeepSeek — not Ollama; a DeepSeek model you pull locally appears under the Local engine instead.
Key actions
Section titled “Key actions”- Chat — send prompts, resume past threads, browse Sessions, read the Telemetry row.
Requires the Model Studio cloud plan key QWEN_CLOUD_KEY in .env.local (from Alibaba Model Studio), then restart the server. One plan feeds both the DeepSeek and Qwen engines. Without it, this engine reads “no cloud plan” — pull a DeepSeek model into Ollama and use the Local engine instead:
ollama pull deepseek-r1:8b- The Qwen and DeepSeek engines share the same
QWEN_CLOUD_KEY.
See also
Section titled “See also”-
Chat — the window this engine answers in.