How to connect DeepSeek Harness to Apily

Requirements

• Windows 10/11, macOS 12+ or a modern 64-bit Linux.
• Node.js — the harness is started through npx. Building from source additionally needs pnpm.
• An Apily API key — a DeepSeek account is not required for a third-party provider.

Install DeepSeek Harness

npx @deepseek-ai/dsh web

The command starts the web UI at http://127.0.0.1:3080 and opens it in the browser. Pass --no-open to keep the server without opening a tab. The project is in developer preview — if the provider form looks different after an update, the connection principle stays the same.

Configuration

1. Open Settings → Models in the web UI and press Add a custom provider.
2. Provider ID in lowercase, for example apily. The id is permanent: a provider can't be renamed, only replaced by a new one. The display name, address, protocol, credential and models stay editable.
3. Base URL: https://apily.pro/v1
4. API protocol: openai-completions — OpenAI Chat Completions. A provider speaks exactly one protocol.
5. API key: apily_your-api-key
6. Under Model catalog press Fetch available models: the request goes to GET /v1/models with the key from the form. Tick the models you want and press Add selected; IDs can also be typed by hand, e.g. gpt-5.6-sol.
7. Save the provider and pick a model in the picker — it becomes the default for new sessions.

Keys are write-only: after saving, the page receives a redacted descriptor rather than the secret itself. The key lives in $DSH_HOME/.credentials.yaml. The same provider can be written straight into $DSH_HOME/settings.yaml — the very document the settings page writes:

llm-pi-ai:
  providers:
    apily:
      apiKeyEnv: APILY_API_KEY
      api: openai-completions
      baseURL: https://apily.pro/v1
      models:
        - id: gpt-5.6-sol

Your first request

curl https://apily.pro/v1/chat/completions \
  -H "Authorization: Bearer apily_your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.6-sol","messages":[{"role":"user","content":"ping"}]}'

Open http://127.0.0.1:3080, pick the model and send a message. A session that has already sent a request keeps its model recorded — the new model applies to new sessions.

Switching models

Models are edited in the provider card: Fetch available models asks the endpoint for the list, and an ID typed by hand works exactly the same. Examples from the Apily catalog:

• gpt-5.6-sol · claude-opus-5 · deepseek-v4-flash-0731 · gemini-3.5-flash

Configuring reasoning

A model typed in by hand declares no reasoning levels, so its Effort entry doesn't appear in the menu and the endpoint alone decides whether the model thinks. Levels are declared with reasoningEfforts in $DSH_HOME/settings.yaml — the key is the menu entry and the value is what travels on the wire as reasoning_effort. Thinking is billed as output tokens.

llm-pi-ai:
  providers:
    apily:
      models:
        - id: gpt-5.6-sol
          reasoningEfforts:
            off: ""
            high: high
            max: max

Common errors and how to fix them

FAQ