Provider configuration

OpenCode supports multiple AI providers, each with their own authentication methods and model offerings. This page documents all supported providers and their configuration requirements.

Supported providers

OpenCode supports the following AI providers (in order of automatic detection priority):

GitHub Copilot - Free with GitHub Copilot subscription
Anthropic - Claude models
OpenAI - GPT models
Google Gemini - Gemini models
Groq - Fast inference for open-source models
OpenRouter - Access to multiple models through one API
AWS Bedrock - Claude models via AWS
Azure OpenAI - OpenAI models via Azure
Google Cloud VertexAI - Gemini models via Google Cloud
xAI - Grok models

Anthropic

Provider for Claude models with industry-leading performance.

Authentication

ANTHROPIC_API_KEY

string

required

Your Anthropic API key. Get one at console.anthropic.com

Configuration

{
  "providers": {
    "anthropic": {
      "apiKey": "sk-ant-api03-...",
      "disabled": false
    }
  }
}

Supported models

Model ID	Name	Context Window	Default Max Tokens	Cost (Input/Output per 1M tokens)	Features
`claude-4-sonnet`	Claude 4 Sonnet	200,000	50,000	$3.00 /$ 15.00	Extended thinking, attachments
`claude-4-opus`	Claude 4 Opus	200,000	4,096	$15.00 /$ 75.00	Attachments
`claude-3.7-sonnet`	Claude 3.7 Sonnet	200,000	50,000	$3.00 /$ 15.00	Extended thinking, attachments
`claude-3.5-sonnet`	Claude 3.5 Sonnet	200,000	5,000	$3.00 /$ 15.00	Attachments
`claude-3.5-haiku`	Claude 3.5 Haiku	200,000	4,096	$0.80 /$ 4.00	Attachments
`claude-3-opus`	Claude 3 Opus	200,000	4,096	$15.00 /$ 75.00	Attachments
`claude-3-haiku`	Claude 3 Haiku	200,000	4,096	$0.25 /$ 1.25	Attachments

Extended thinking models support the reasoningEffort parameter for deeper analysis.

OpenAI

Provider for GPT models including the reasoning-capable o-series.

Authentication

OPENAI_API_KEY

string

required

Your OpenAI API key. Get one at platform.openai.com

Configuration

{
  "providers": {
    "openai": {
      "apiKey": "sk-...",
      "disabled": false
    }
  }
}

Supported models

Model ID	Name	Context Window	Default Max Tokens	Cost (Input/Output per 1M tokens)	Features
`gpt-4.1`	GPT 4.1	1,047,576	20,000	$2.00 /$ 8.00	Attachments, prompt caching
`gpt-4.1-mini`	GPT 4.1 Mini	200,000	20,000	$0.40 /$ 1.60	Attachments, prompt caching
`gpt-4.1-nano`	GPT 4.1 Nano	1,047,576	20,000	$0.10 /$ 0.40	Attachments, prompt caching
`gpt-4.5-preview`	GPT 4.5 Preview	128,000	15,000	$75.00 /$ 150.00	Attachments, prompt caching
`gpt-4o`	GPT-4o	128,000	4,096	$2.50 /$ 10.00	Attachments, prompt caching
`gpt-4o-mini`	GPT-4o Mini	128,000	-	$0.15 /$ 0.60	Attachments, prompt caching
`o1`	o1	200,000	50,000	$15.00 /$ 60.00	Reasoning, attachments, prompt caching
`o1-pro`	o1 Pro	200,000	50,000	$150.00 /$ 600.00	Reasoning, attachments
`o1-mini`	o1 Mini	128,000	50,000	$1.10 /$ 4.40	Reasoning, attachments, prompt caching
`o3`	o3	200,000	-	$10.00 /$ 40.00	Reasoning, attachments, prompt caching
`o3-mini`	o3 Mini	200,000	50,000	$1.10 /$ 4.40	Reasoning, prompt caching
`o4-mini`	o4 Mini	128,000	50,000	$1.10 /$ 4.40	Reasoning, attachments, prompt caching

Reasoning models (o-series) support the reasoningEffort parameter: low, medium, or high.

Google Gemini

Provider for Gemini models with large context windows.

Authentication

GEMINI_API_KEY

string

required

Your Google AI Studio API key. Get one at aistudio.google.com

Configuration

{
  "providers": {
    "gemini": {
      "apiKey": "AI...",
      "disabled": false
    }
  }
}

Supported models

Model ID	Name	Context Window	Default Max Tokens	Cost (Input/Output per 1M tokens)	Features
`gemini-2.5`	Gemini 2.5 Pro	1,000,000	50,000	$1.25 /$ 10.00	Attachments
`gemini-2.5-flash`	Gemini 2.5 Flash	1,000,000	50,000	$0.15 /$ 0.60	Attachments
`gemini-2.0-flash`	Gemini 2.0 Flash	1,000,000	6,000	$0.10 /$ 0.40	Attachments
`gemini-2.0-flash-lite`	Gemini 2.0 Flash Lite	1,000,000	6,000	$0.05 /$ 0.30	Attachments

Groq

Provider for fast inference with open-source models.

Authentication

GROQ_API_KEY

string

required

Your Groq API key. Get one at console.groq.com

Configuration

{
  "providers": {
    "groq": {
      "apiKey": "gsk_...",
      "disabled": false
    }
  }
}

Supported models

Model ID	Name	Context Window	Cost (Input/Output per 1M tokens)
`qwen-qwq`	Qwen QwQ 32B	128,000	$0.29 /$ 0.39
`llama-3.3-70b-versatile`	Llama 3.3 70B Versatile	128,000	$0.59 /$ 0.79
`meta-llama/llama-4-scout-17b-16e-instruct`	Llama 4 Scout	128,000	$0.11 /$ 0.34
`meta-llama/llama-4-maverick-17b-128e-instruct`	Llama 4 Maverick	128,000	$0.20 /$ 0.20
`deepseek-r1-distill-llama-70b`	DeepSeek R1 Distill	128,000	$0.75 /$ 0.99

OpenRouter

Provider for accessing multiple models through a single API.

Authentication

OPENROUTER_API_KEY

string

required

Your OpenRouter API key. Get one at openrouter.ai/keys

Configuration

{
  "providers": {
    "openrouter": {
      "apiKey": "sk-or-...",
      "disabled": false
    }
  }
}

Supported models

OpenRouter provides access to models from multiple providers. Prefix model IDs with openrouter.:

openrouter.gpt-4.1, openrouter.gpt-4.1-mini, openrouter.gpt-4.1-nano
openrouter.o1, openrouter.o1-mini, openrouter.o3, openrouter.o3-mini, openrouter.o4-mini
openrouter.claude-3.7-sonnet, openrouter.claude-3.5-sonnet, openrouter.claude-3.5-haiku
openrouter.gemini-2.5, openrouter.gemini-2.5-flash
openrouter.deepseek-r1-free - Free access to DeepSeek R1

Pricing matches the underlying provider. See OpenRouter’s model pricing for details.

AWS Bedrock

Provider for Claude models via AWS infrastructure.

Authentication

Bedrock uses AWS credentials. OpenCode detects credentials in this order:

AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables
AWS_PROFILE or AWS_DEFAULT_PROFILE environment variables
EC2 instance profiles or ECS task roles

AWS_REGION

string

default:"us-east-1"

AWS region for Bedrock (e.g., us-east-1, us-west-2)

Configuration

{
  "providers": {
    "bedrock": {
      "disabled": false
    }
  }
}

Supported models

Model ID	Name	Cost (Input/Output per 1M tokens)
`bedrock.claude-3.7-sonnet`	Claude 3.7 Sonnet	$3.00 /$ 15.00

Bedrock models automatically disable prompt caching as it’s not supported in the Bedrock API.

Azure OpenAI

Provider for OpenAI models via Azure infrastructure.

Authentication

Azure OpenAI supports two authentication methods: API Key authentication:

AZURE_OPENAI_ENDPOINT

string

required

Your Azure OpenAI endpoint (e.g., https://your-resource.openai.azure.com)

AZURE_OPENAI_API_KEY

string

Your Azure OpenAI API key (optional when using Entra ID)

AZURE_OPENAI_API_VERSION

string

default:"2025-04-01-preview"

Azure OpenAI API version

Entra ID authentication: If AZURE_OPENAI_API_KEY is not provided, OpenCode uses Azure’s DefaultAzureCredential for authentication.

Configuration

{
  "providers": {
    "azure": {
      "apiKey": "your-key",
      "disabled": false
    }
  }
}

Supported models

Azure OpenAI supports the same models as OpenAI. Prefix model IDs with azure.:

azure.gpt-4.1, azure.gpt-4.1-mini, azure.gpt-4.1-nano
azure.gpt-4o, azure.gpt-4o-mini
azure.o1, azure.o1-mini, azure.o3, azure.o3-mini, azure.o4-mini

Pricing is similar to OpenAI’s standard pricing but may vary by region.

Google Cloud VertexAI

Provider for Gemini models via Google Cloud infrastructure.

Authentication

VertexAI uses Google Cloud credentials. OpenCode detects credentials when these environment variables are set:

VERTEXAI_PROJECT

string

required

Your Google Cloud project ID

VERTEXAI_LOCATION

string

required

Google Cloud region (e.g., us-central1)

Alternatively:

GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_REGION (or GOOGLE_CLOUD_LOCATION)

Configuration

{
  "providers": {
    "vertexai": {
      "disabled": false
    }
  }
}

Supported models

| Model ID | Name | Context Window | Default Max Tokens | |----------|------|----------------|--------------------|---| | vertexai.gemini-2.5 | Gemini 2.5 Pro | 1,000,000 | 50,000 | | vertexai.gemini-2.5-flash | Gemini 2.5 Flash | 1,000,000 | 50,000 |

GitHub Copilot

Provider for models via GitHub Copilot subscription. Includes access to multiple models at no additional cost.

Authentication

GitHub Copilot authentication is auto-detected from:

GITHUB_TOKEN environment variable
GitHub Copilot CLI configuration:
- ~/.config/github-copilot/hosts.json (Linux/macOS)
- ~/.config/github-copilot/apps.json (Linux/macOS)
- %LOCALAPPDATA%\github-copilot\ (Windows)

Configuration

{
  "providers": {
    "copilot": {
      "disabled": false
    }
  }
}

Supported models

All models are included with GitHub Copilot subscription at $0 cost:

Model ID	Name	Context Window	Default Max Tokens	Features
`copilot.gpt-4.1`	GPT 4.1	128,000	16,384	Reasoning, attachments
`copilot.gpt-4o`	GPT-4o	128,000	16,384	Attachments
`copilot.gpt-4o-mini`	GPT-4o Mini	128,000	4,096	Attachments
`copilot.claude-3.7-sonnet`	Claude 3.7 Sonnet	200,000	16,384	Attachments
`copilot.claude-sonnet-4`	Claude Sonnet 4	128,000	16,000	Attachments
`copilot.o1`	o1	200,000	100,000	Reasoning
`copilot.o3-mini`	o3-mini	200,000	100,000	Reasoning
`copilot.o4-mini`	o4-mini	128,000	16,384	Reasoning, attachments
`copilot.gemini-2.5-pro`	Gemini 2.5 Pro	128,000	64,000	Attachments
`copilot.gemini-2.0-flash`	Gemini 2.0 Flash	1,000,000	8,192	Attachments

GitHub Copilot provides the best value with access to multiple providers at no additional cost beyond the subscription.

xAI

Provider for Grok models.

Authentication

XAI_API_KEY

string

required

Your xAI API key

Configuration

{
  "providers": {
    "xai": {
      "apiKey": "xai-...",
      "disabled": false
    }
  }
}

Supported models

| Model ID | Name | Context Window | Default Max Tokens | Cost (Input/Output per 1M tokens) | |----------|------|----------------|--------------------|------------------------------------|---| | grok-3-beta | Grok 3 Beta | 131,072 | 20,000 |

3.00 /

15.00 | | grok-3-mini-beta | Grok 3 Mini Beta | 131,072 | 20,000 |

0.30 /

0.50 | | grok-3-fast-beta | Grok 3 Fast Beta | 131,072 | 20,000 |

5.00 /

25.00 | | grok-3-mini-fast-beta | Grok 3 Mini Fast Beta | 131,072 | 20,000 |

0.60 /

4.00 |

Default model selection

OpenCode automatically selects default models based on available providers in this priority order:

GitHub Copilot → copilot.gpt-4o
Anthropic → claude-4-sonnet
OpenAI → gpt-4.1 (coder), gpt-4.1-mini (task/title)
Gemini → gemini-2.5 (coder), gemini-2.5-flash (task/title)
Groq → qwen-qwq
OpenRouter → openrouter.claude-3.7-sonnet
xAI → grok-3-beta
AWS Bedrock → bedrock.claude-3.7-sonnet
Azure OpenAI → azure.gpt-4.1
VertexAI → vertexai.gemini-2.5

You can override defaults by explicitly configuring the agents section in your configuration file.

CLI Reference

Configuration

​Supported providers

​Anthropic

​Authentication

​Configuration

​Supported models

​OpenAI

​Authentication

​Configuration

​Supported models

​Google Gemini

​Authentication

​Configuration

​Supported models

​Groq

​Authentication

​Configuration

​Supported models

​OpenRouter

​Authentication

​Configuration

​Supported models

​AWS Bedrock

​Authentication

​Configuration

​Supported models

​Azure OpenAI

​Authentication

​Configuration

​Supported models

​Google Cloud VertexAI

​Authentication

​Configuration

​Supported models

​GitHub Copilot

​Authentication

​Configuration

​Supported models

​xAI

​Authentication

​Configuration

​Supported models

​Default model selection

Build docs developers (and LLMs) love

Supported providers

Anthropic

Authentication

Configuration

Supported models

OpenAI

Authentication

Configuration

Supported models

Google Gemini

Authentication

Configuration

Supported models

Groq

Authentication

Configuration

Supported models

OpenRouter

Authentication

Configuration

Supported models

AWS Bedrock

Authentication

Configuration

Supported models

Azure OpenAI

Authentication

Configuration

Supported models

Google Cloud VertexAI

Authentication

Configuration

Supported models

GitHub Copilot

Authentication

Configuration

Supported models

xAI

Authentication

Configuration

Supported models

Default model selection