Silicon Worker

English

简体中文

English

简体中文

Appearance

Configure Model Providers ​

Silicon Worker has no built-in LLM; it connects to a model service you choose. This guide helps you connect different kinds of providers and set up primary/fallback models for better reliability.

If you just want to connect a model quickly, see Connect Your First Model in Get Started; this page goes deeper.

Supported provider types ​

Silicon Worker connects to models through two protocol families, and most services fall into one of them:

  • OpenAI-compatible (default): any service exposing /v1/chat/completions with Bearer Token auth. OpenAI, Azure OpenAI, DeepSeek, Qwen, Zhipu, SiliconFlow, and local Ollama all fall into this category.
  • Anthropic native: Claude models, using the /v1/messages endpoint with x-api-key auth.

Add a provider ​

  1. Open Silicon Worker → Settings → Models / Providers.
  2. Choose the provider type (OpenAI-compatible / Anthropic).
  3. Enter your API Key.
  4. For custom or regional services, also set the Base URL and model name.
  5. Save and select this model as the active one.

What to enter

Use the official documentation of your service for the Base URL and model name. OpenAI-compatible services usually give an endpoint like https://.../v1; the model name must match the server's identifier exactly.

Chinese providers example ​

DeepSeek, Qwen, Zhipu, SiliconFlow, and others are typically OpenAI-compatible. The setup is the same:

  1. Choose the "OpenAI-compatible" provider type.
  2. Set the Base URL to the service's endpoint.
  3. Set the model name to the specific model identifier you want to use.
  4. Enter the API Key issued by that service.

Primary & fallback with failover ​

You can configure a primary model and a fallback model for a session. If the primary call fails (e.g. network error, rate limit), Silicon Worker automatically switches to the fallback to keep the task running, improving reliability.

  1. In Settings → Models / Providers, make sure at least two usable models are configured.
  2. Select the primary model.
  3. Designate the fallback model.

Suggestion

Setting the primary and fallback to models from different providers reduces the impact of any single service's outages.

Result and verification ​

Once configured, return to the chat and give it a simple task to confirm the model responds. If the primary model hits a transient failure, you'll see it switch to the fallback automatically.

About API Keys

API Keys are stored locally on your device. Keep them safe and don't share them.