Providers Overview

Learn about AI providers supported by Forge and how to configure them

Forge supports multiple AI providers, allowing you to choose the model and service that best fits your needs. You can configure providers using the interactive login command or through environment variables (deprecated).

Quick Start

The easiest way to configure a provider is using the interactive login flow:

forge provider login

This command will:

• Display a list of available providers
• Guide you through entering the required credentials
• Securely store your credentials locally

Managing Providers

List Available Providers

forge provider list

Add or Update Provider Credentials

forge provider login

Remove Provider Credentials

forge provider logout

Supported Providers

Forge supports a wide range of AI providers:

Cloud Providers

• OpenAI - GPT-4, GPT-3.5, and O-series models
• Anthropic/Claude - Claude Sonnet, Opus, and Haiku models
• OpenRouter - Access multiple models through a single API
• Google Vertex AI - Gemini and Claude models on Google Cloud
• AWS Bedrock - Claude, Llama, and other models on AWS

OpenAI-Compatible Providers

• Cerebras
• XAI (Grok)
• Z.AI (GLM models)
• DeepSeek
• IO Intelligence
• Requesty

Custom Providers

• Custom OpenAI-Compatible - Any OpenAI-compatible API
• Custom Anthropic-Compatible - Any Anthropic-compatible API
• Local inference engines (Ollama, LM Studio, llama.cpp, vLLM)

Provider Configuration

Using .env files for provider configuration is deprecated and will be removed in a future version. Please use forge provider login instead.

Each provider requires:

1. API Key - Your authentication credentials
2. URL Parameters (optional) - Some providers require additional parameters like project IDs or regions

Credential Storage

Credentials are stored securely in:

• Location: ~/.config/forge/credentials.json (Linux/macOS) or %APPDATA%\forge\credentials.json (Windows)
• Format: JSON with API keys and configuration parameters
• Security: File permissions are set to be readable only by the owner

Selecting a Model

After configuring a provider, you can:

1. Set a default model in forge.yaml:

   model: claude-sonnet-4-20250514
   

2. Switch models during a session using the /model command:

   /model gpt-4o
   

3. List available models for your configured providers:

   /model
   

Migration from Environment Variables

If you have existing credentials in environment variables, Forge will automatically migrate them to file-based storage on first run. You’ll see a message indicating which providers were migrated.

Next Steps

• Configure OpenRouter for access to multiple models
• Set up Anthropic/Claude for advanced reasoning
• Configure OpenAI for GPT models
• Use Google Vertex AI for Gemini models
• Set up AWS Bedrock for enterprise deployments