Provider Configuration

Configure LLM providers and models in Crush

Crush supports multiple LLM providers including Anthropic, OpenAI, Google Gemini, AWS Bedrock, Azure, and many more. You can configure providers using environment variables or configuration files.

Quick Start with Environment Variables

The quickest way to get started is to set an API key for your preferred provider. Crush recognizes these environment variables:

Environment Variable	Provider
ANTHROPIC_API_KEY	Anthropic
OPENAI_API_KEY	OpenAI
VERCEL_API_KEY	Vercel AI Gateway
GEMINI_API_KEY	Google Gemini
SYNTHETIC_API_KEY	Synthetic
ZAI_API_KEY	Z.ai
MINIMAX_API_KEY	MiniMax
HF_TOKEN	Hugging Face Inference
CEREBRAS_API_KEY	Cerebras
OPENROUTER_API_KEY	OpenRouter
IONET_API_KEY	io.net
GROQ_API_KEY	Groq
VERTEXAI_PROJECT	Google Cloud VertexAI (Gemini)
VERTEXAI_LOCATION	Google Cloud VertexAI (Gemini)
AWS_ACCESS_KEY_ID	Amazon Bedrock (Claude)
AWS_SECRET_ACCESS_KEY	Amazon Bedrock (Claude)
AWS_REGION	Amazon Bedrock (Claude)
AWS_PROFILE	Amazon Bedrock (Custom Profile)
AWS_BEARER_TOKEN_BEDROCK	Amazon Bedrock
AZURE_OPENAI_API_ENDPOINT	Azure OpenAI models
AZURE_OPENAI_API_KEY	Azure OpenAI models (optional when using Entra ID)
AZURE_OPENAI_API_VERSION	Azure OpenAI models

Just set an environment variable and run Crush. If no API key is found, Crush will prompt you to enter one.

Custom Provider Configuration

For advanced scenarios, you can configure custom providers in your crush.json file.

OpenAI-Compatible Providers

Crush supports two OpenAI provider types:

• openai - For proxying/routing requests through OpenAI
• openai-compat - For non-OpenAI providers with OpenAI-compatible APIs

Here’s an example configuration for Deepseek:

{
  "$schema": "https://charm.land/crush.json",
  "providers": {
    "deepseek": {
      "type": "openai-compat",
      "base_url": "https://api.deepseek.com/v1",
      "api_key": "$DEEPSEEK_API_KEY",
      "models": [
        {
          "id": "deepseek-chat",
          "name": "Deepseek V3",
          "cost_per_1m_in": 0.27,
          "cost_per_1m_out": 1.1,
          "cost_per_1m_in_cached": 0.07,
          "cost_per_1m_out_cached": 1.1,
          "context_window": 64000,
          "default_max_tokens": 5000
        }
      ]
    }
  }
}

Anthropic-Compatible Providers

For providers that use the Anthropic API format:

{
  "$schema": "https://charm.land/crush.json",
  "providers": {
    "custom-anthropic": {
      "type": "anthropic",
      "base_url": "https://api.anthropic.com/v1",
      "api_key": "$ANTHROPIC_API_KEY",
      "extra_headers": {
        "anthropic-version": "2023-06-01"
      },
      "models": [
        {
          "id": "claude-sonnet-4-20250514",
          "name": "Claude Sonnet 4",
          "cost_per_1m_in": 3,
          "cost_per_1m_out": 15,
          "cost_per_1m_in_cached": 3.75,
          "cost_per_1m_out_cached": 0.3,
          "context_window": 200000,
          "default_max_tokens": 50000,
          "can_reason": true,
          "supports_attachments": true
        }
      ]
    }
  }
}

Local Models

You can run Crush with local models using OpenAI-compatible servers.

Ollama

{
  "providers": {
    "ollama": {
      "name": "Ollama",
      "base_url": "http://localhost:11434/v1/",
      "type": "openai-compat",
      "models": [
        {
          "name": "Qwen 3 30B",
          "id": "qwen3:30b",
          "context_window": 256000,
          "default_max_tokens": 20000
        }
      ]
    }
  }
}

LM Studio

{
  "providers": {
    "lmstudio": {
      "name": "LM Studio",
      "base_url": "http://localhost:1234/v1/",
      "type": "openai-compat",
      "models": [
        {
          "name": "Qwen 3 30B",
          "id": "qwen/qwen3-30b-a3b-2507",
          "context_window": 256000,
          "default_max_tokens": 20000
        }
      ]
    }
  }
}

Cloud Providers

Amazon Bedrock

Crush supports running Anthropic models through Bedrock (caching disabled):

1. Configure AWS credentials: aws configure
2. Set AWS_REGION or AWS_DEFAULT_REGION environment variable
3. (Optional) Use a specific profile: AWS_PROFILE=myprofile crush
4. Alternative: Set AWS_BEARER_TOKEN_BEDROCK instead of running aws configure

Vertex AI Platform

Vertex AI appears when VERTEXAI_PROJECT and VERTEXAI_LOCATION are set:

gcloud auth application-default login

Configure specific models:

{
  "$schema": "https://charm.land/crush.json",
  "providers": {
    "vertexai": {
      "models": [
        {
          "id": "claude-sonnet-4@20250514",
          "name": "VertexAI Sonnet 4",
          "cost_per_1m_in": 3,
          "cost_per_1m_out": 15,
          "cost_per_1m_in_cached": 3.75,
          "cost_per_1m_out_cached": 0.3,
          "context_window": 200000,
          "default_max_tokens": 50000,
          "can_reason": true,
          "supports_attachments": true
        }
      ]
    }
  }
}

Model Configuration Options

When defining models in your provider configuration, you can specify:

• id (required) - Model identifier used by the provider API
• name (required) - Human-readable model name
• cost_per_1m_in - Cost per 1M input tokens
• cost_per_1m_out - Cost per 1M output tokens
• cost_per_1m_in_cached - Cost per 1M cached input tokens
• cost_per_1m_out_cached - Cost per 1M cached output tokens
• context_window - Maximum context window size
• default_max_tokens - Default maximum tokens for responses
• can_reason - Whether the model supports extended reasoning
• supports_attachments - Whether the model supports file attachments

Provider Auto-Updates

By default, Crush automatically updates the provider database from Catwalk, the open source Crush provider database.

Disabling Auto-Updates

For air-gapped or restricted environments:

{
  "$schema": "https://charm.land/crush.json",
  "options": {
    "disable_provider_auto_update": true
  }
}

Or use the environment variable:

export CRUSH_DISABLE_PROVIDER_AUTO_UPDATE=1

Manual Updates

Update providers manually:

# Update from Catwalk
crush update-providers

# Update from custom URL
crush update-providers https://example.com/

# Update from local file
crush update-providers /path/to/local-providers.json

# Reset to embedded version
crush update-providers embedded

Environment Variable Expansion

In configuration files, you can reference environment variables using the $VARIABLE_NAME syntax:

{
  "providers": {
    "openai": {
      "api_key": "$OPENAI_API_KEY"
    }
  }
}

Crush automatically expands these variables when loading the configuration.

Next Steps

Model Selection

Learn how to list and select models

Permissions

Configure tool permissions