DeepSeek V4 Pro in DeepSeek Harness (dsh): deepseek-v4-pro-free Isn't Available — Use deepseek-v4-pro or Free V4 Flash

Quick Answer

deepseek-v4-pro-free is not available. TeamoRouter has no model with that ID, so dsh sessions set to it fail with "Model 'deepseek-v4-pro-free' is not available". In DeepSeek Harness, use:

• deepseek-v4-pro for DeepSeek V4 Pro. It's a paid model (model page).
• deepseek-v4-flash-free for the lower-priced V4 Flash tier, still billed (model page).

The old three environment variables don't select these models in dsh 0.2.x: DEEPSEEK_DEFAULT_MODEL is ignored, and the built-in DeepSeek provider only knows DeepSeek's own model names. Add TeamoRouter as a provider instead.

New accounts start with a $0 balance — top up before your first request. Top up from $5. If a request returns 402 (400 on the OpenAI format) with "wallet balance is insufficient", your balance is empty — your base URL and key are fine. Top up and retry.

Set up both models in dsh

1. Create an API key in the TeamoRouter console.
2. Save this as ~/.dsh/settings.yaml:

llm-pi-ai:
  providers:
    teamorouter:
      displayName: TeamoRouter
      apiKeyEnv: TEAMOROUTER_API_KEY
      api: openai-completions
      baseURL: https://api.agentapiai.com/v1
      models:
        - id: deepseek-v4-flash-free
          name: DeepSeek V4 Flash Free
        - id: deepseek-v4-pro
          name: DeepSeek V4 Pro
agent-default-model:
  provider: teamorouter
  model: deepseek-v4-flash-free

3. Put the key in ~/.dsh/.env:

echo 'TEAMOROUTER_API_KEY=sk-teamo-your-key' >> ~/.dsh/.env
chmod 600 ~/.dsh/.env

4. Run npx @deepseek-ai/dsh web, open http://127.0.0.1:3080 and switch between the two models in the model picker.

On first start dsh 0.2.x imports settings.yaml into the profile you start and renames it to settings.yaml.imported; other profiles need --patch. Details: Connect DeepSeek Harness to TeamoRouter. We tested this config with dsh 0.2.1-rc.3 on November 2, 2026: dsh sent deepseek-v4-flash-free to /v1/chat/completions, and the request reached the TeamoRouter gateway.

Route by difficulty

With both models in the list, a sensible split is:

• deepseek-v4-flash-free for volume work: explaining a function, commit messages, single-file edits, quick lookups.
• deepseek-v4-pro for multi-file reasoning, long agent loops and anything where a wrong answer costs you a redo.

In an agent loop each tool round-trip is a separate model request: the model asks to read a file, the tool returns it, the model asks for the next step. A multi-file refactor can take 10 to 15 requests, so give the agent a clear task with the files it needs and stop it when it starts going in circles.

Price

DeepSeek's official price for V4 Pro is $1.32 per million input tokens and $3.96 per million output tokens at peak, half that off-peak (DeepSeek pricing, November 2, 2026). TeamoRouter's current prices are on the pricing page. No subscription; top up from $5 by card or with USDT, USDC or ETH (how to pay with USDT).

FAQ

Is there a free V4 Pro in dsh?

No. The lower-priced DeepSeek tier on TeamoRouter is V4 Flash, deepseek-v4-flash-free, which is billed at a lower rate, not free; V4 Pro is the paid deepseek-v4-pro.

Where do I find details on the lower-priced tier?

On the deepseek-v4-flash-free model page.

My dsh session says the model is not available. What do I change?

Replace deepseek-v4-pro-free with deepseek-v4-pro or deepseek-v4-flash-free in settings.yaml (or in the web profile's cordis.patch.yml if it was already imported), then restart dsh.

Can I keep using the environment variables?

DEEPSEEK_API_KEY and DEEPSEEK_BASE_URL still reach TeamoRouter, but dsh then uses its built-in DeepSeek provider with deepseek-flash or deepseek-v4-pro only, not deepseek-v4-flash-free. For the lower-priced -free tier, use the provider config above.