How do I use DeepSeek V4 Flash in Roo Code with CloudGPU's API?

In Roo Code, choose the “OpenAI Compatible” provider, set the Base URL to https://cloudgpu.app/v1, paste your CloudGPU key and enter the model ID deepseek-v4-flash. Roo Code’s docs describe this OpenAI-compatible setup. Roo Code only works with models that support native tool calling, so run the tool-call check below first.

Updated 2026-10-27

What you need

• A CloudGPU API key (cgw-sk-...) from the API Console. Create one from /api; it’s shown once. The commands below read it from CLOUDGPU_API_KEY.
• The model ID deepseek-v4-flash, as listed on live /api (checked 2026-10-27).
• Roo Code installed (follow its official install docs).

Setup (from Roo Code docs, “OpenAI Compatible”)

1. Open the Roo Code settings panel.
2. API Provider: select “OpenAI Compatible”.
3. Base URL: https://cloudgpu.app/v1.
4. API Key: your CloudGPU key (cgw-sk-...).
5. Model: enter deepseek-v4-flash.
6. Model Configuration (Max Output Tokens, Context Window and so on) is optional. We don’t list values for deepseek-v4-flash here; leave the defaults unless you know the limits you need.

Roo Code field	Value
API Provider	OpenAI Compatible
Base URL	https://cloudgpu.app/v1
API Key	cgw-sk-... (from the API Console)
Model	deepseek-v4-flash

Check it works

Before opening Roo Code, test the key and model ID with the request from /api/docs:

curl https://cloudgpu.app/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $CLOUDGPU_API_KEY" \
  -d '{"model":"deepseek-v4-flash","messages":[{"role":"user","content":"Say hi in one sentence."}]}'

GET https://cloudgpu.app/v1/models lists the model IDs your key can use.

Roo Code’s docs say it uses native tool calling only: “If a model does not support native tool calling, it cannot be used with Roo Code.” We haven’t confirmed tool calling for deepseek-v4-flash inside Roo Code, so check that the model accepts the tools parameter:

curl https://cloudgpu.app/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $CLOUDGPU_API_KEY" \
  -d '{"model":"deepseek-v4-flash","messages":[{"role":"user","content":"What is the weather in Paris? Use the tool."}],"tools":[{"type":"function","function":{"name":"get_weather","description":"Get the weather for a city","parameters":{"type":"object","properties":{"city":{"type":"string"}},"required":["city"]}}}]}'

If the reply has a tool_calls entry, the model accepted the tool. If you get an error or a plain text answer instead, Roo Code is unlikely to work with that model; try another model from /v1/models.

Troubleshooting

• “Invalid API Key” (HTTP 401 invalid_api_key): the key is wrong or revoked; create a new one in the console (keys are shown once).
• “Model Not Found” (HTTP 400 model_not_found): check the ID against /v1/models.
• Connection errors: check the Base URL is exactly https://cloudgpu.app/v1.
• Tool-calling errors: Roo Code needs native tool calling. Run the tool-call check above; if it fails, switch to a model that passes it.

FAQ

Does Roo Code work with an OpenAI-compatible API?

Yes. Roo Code’s docs describe connecting to any OpenAI-compatible endpoint with a base URL, an API key and a model ID.

What model ID do I use for DeepSeek V4 Flash?

deepseek-v4-flash, exactly as listed on /api and returned by /v1/models.

Does deepseek-v4-flash support the tool calling Roo Code needs?

We haven’t confirmed it inside Roo Code. Roo Code only works with models that support native tool calling, so run the tool-call check on this page before you rely on it.

Where do I see what it costs?

See current rates on /api.

Sources

• Roo Code docs, “Using OpenAI Compatible Providers With Roo Code”, https://docs.roocode.com/providers/openai-compatible (read 2026-10-27)
• CloudGPU /api and /api/docs, live (read 2026-10-27)