Opencode Providers

OpenCode Providers

> 📚 Official Docs: For the latest information, always refer to the official documentation: > https://opencode.ai/docs/providers/

Overview

OpenCode uses the AI SDK and Models.dev to support 75+ LLM providers and local models. Providers are configured through opencode.json and authenticated via the /connect command.

Quick start:

1. Run /connect in the TUI and select your provider
2. Enter your API key when prompted
3. Run /models to select a model

Credentials

API keys added via /connect are stored in ~/.local/share/opencode/auth.json.

Provider Config

Customize providers through the provider section in opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "anthropic": {
      "options": {
        "baseURL": "https://api.anthropic.com/v1"
      }
    }
  }
}

Key config fields:

• npm — AI SDK package (e.g. @ai-sdk/openai-compatible for OpenAI-compatible endpoints, @ai-sdk/openai for /v1/responses endpoints)
• name — Display name in UI
• options.baseURL — API endpoint URL
• options.apiKey — API key (or use /connect)
• options.headers — Custom headers sent with each request
• models — Map of model IDs to configurations
• limit.context — Maximum input tokens the model accepts
• limit.output — Maximum tokens the model can generate

---

OpenCode Zen

OpenCode Zen is a curated list of tested and verified models provided by the OpenCode team. It works like any other provider — pay-as-you-go, completely optional.

Setup:

1. Sign in at opencode.ai/auth, add billing details, copy API key
2. Run /connect, select OpenCode Zen, paste API key
3. Run /models to see recommended models

Free models available:

• DeepSeek V4 Flash Free
• MiMo-V2.5 Free
• Nemotron 3 Super Free
• Big Pickle (stealth model)

Pricing (per 1M tokens, selected models):

| Model | Input | Output | Cached Read | |-------|-------|--------|-------------| | GPT 5.5 (≤272K) | $5.00 | $30.00 | $0.50 | | GPT 5.4 | $2.50 | $15.00 | $0.25 | | Claude Opus 4.8 | $5.00 | $25.00 | $0.50 | | Claude Sonnet 4.6 | $3.00 | $15.00 | $0.30 | | Gemini 3.5 Flash | $1.50 | $9.00 | $0.15 | | Kimi K2.5 | $0.60 | $3.00 | $0.10 | | MiniMax M2.7 | $0.30 | $1.20 | $0.06 | | Qwen3.6 Plus | $0.20 | $1.20 | $0.02 | | GPT 5 Nano | $0.05 | $0.40 | $0.005 |

Features:

• Auto-reload: Balance below $5 triggers $20 reload (configurable, can disable)
• Monthly limits: Set per-workspace and per-member spending caps
• Teams: Invite teammates, assign roles (Admin/Member), curate model access, bring your own keys
• Privacy: Zero-retention by default; free models may use data for improvement; OpenAI/Anthropic retain requests 30 days

Model ID format in config: opencode/ (e.g. opencode/gpt-5.5)

Endpoints:

• OpenAI models: https://opencode.ai/zen/v1/responses
• Anthropic models: https://opencode.ai/zen/v1/messages
• Google models: https://opencode.ai/zen/v1/models/
• Other models: https://opencode.ai/zen/v1/chat/completions

---

OpenCode Go

OpenCode Go is a low-cost subscription plan for popular open coding models.

• $5 first month, $10/month after that
• Models tested and verified for coding tasks

Setup:

1. Run /connect, select OpenCode Go, go to opencode.ai/auth
2. Sign in, add billing, copy API key
3. Paste API key in terminal
4. Run /models to see available models

Usage limits:

• 5 hour limit: $12
• Weekly limit: $30
• Monthly limit: $60

Available models:

• GLM-5
• GLM-5.1
• Kimi K2.5
• Kimi K2.6
• MiMo-V2.5
• MiMo-V2.5-Pro
• MiniMax M2.5
• MiniMax M2.7
• Qwen3.6 Plus
• Qwen3.7 Max
• DeepSeek V4 Pro
• DeepSeek V4 Flash

---

Major Providers

Anthropic (Claude)

Supports Claude Pro/Max OAuth and API keys.

Setup:

1. Run /connect, select Anthropic
2. Select Claude Pro/Pro Max for OAuth (opens browser) or Manually enter API Key
3. Run /models to select model

Available models: Claude Opus 4, Claude Sonnet 4, Claude Haiku 3.5, and others.

Config example:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "anthropic": {
      "options": {
        "baseURL": "https://api.anthropic.com/v1"
      }
    }
  }
}

Note: Claude Pro/Max OAuth is the only supported way to use Anthropic subscriptions. Third-party plugins that use Claude subscriptions are explicitly prohibited by Anthropic and are no longer bundled with OpenCode as of v1.3.0.

OpenAI

Supports ChatGPT Plus/Pro OAuth and API keys.

Setup:

1. Run /connect, select OpenAI
2. Select ChatGPT Plus/Pro for OAuth or Manually enter API Key
3. Run /models to select model

Using API keys: Select Manually enter API Key and paste your key.

xAI

Supports SuperGrok OAuth, device-code flow, and API keys.

Auth Method 1 — SuperGrok OAuth (browser):

1. Run /connect, select xAI
2. Select xAI Grok OAuth (SuperGrok Subscription)
3. Browser opens for consent

Auth Method 2 — SuperGrok device-code (headless/VPS):

1. Run /connect, select xAI
2. Select xAI Grok OAuth (Headless/Remote/VPS)
3. Visit the verification URL on another device and enter the displayed code

Auth Method 3 — API key:

1. Create API key at console.x.ai
2. Run /connect, select xAI
3. Select Manually enter API Key and paste your key

Available models: Grok Beta and others.

GitHub Copilot

Uses device code flow. Requires GitHub Copilot subscription (some models need Pro+).

Setup:

1. Run /connect, search for GitHub Copilot
2. Navigate to github.com/login/device and enter the displayed code
3. Wait for authorization
4. Run /models to select model

┌ Login with GitHub Copilot
│
│ https://github.com/login/device
│
│ Enter code: 8F43-6FCF
│
└ Waiting for authorization...

Google Vertex AI

Requires Google Cloud project with Vertex AI API enabled.

Environment variables:

export GOOGLE_CLOUD_PROJECT=your-project-id
export VERTEX_LOCATION=global  # optional, defaults to global
export GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account.json

Or authenticate with gcloud:

gcloud auth application-default login

Config example:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "google-vertex": {
      "options": {
        "baseURL": "https://global-aiplatform.googleapis.com"
      }
    }
  }
}

Tip: The global region improves availability at no extra cost. Use regional endpoints (e.g. us-central1) for data residency requirements.

Amazon Bedrock

Supports multiple authentication methods with configurable precedence.

Environment variables:

# Option 1: AWS access keys
export AWS_ACCESS_KEY_ID=XXX
export AWS_SECRET_ACCESS_KEY=YYY

# Option 2: Named AWS profile
export AWS_PROFILE=my-profile

# Option 3: Bedrock bearer token
export AWS_BEARER_TOKEN_BEDROCK=XXX

Config example:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "amazon-bedrock": {
      "options": {
        "region": "us-east-1",
        "profile": "my-aws-profile",
        "endpoint": "https://bedrock-runtime.us-east-1.vpce-xxxxx.amazonaws.com"
      }
    }
  }
}

Available options:

• region — AWS region (e.g. us-east-1, eu-west-1)
• profile — Named profile from ~/.aws/credentials
• endpoint — Custom endpoint URL for VPC endpoints (alias for baseURL)

Authentication methods:

• AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY — IAM user access keys
• AWS_PROFILE — Named profiles from ~/.aws/credentials
• AWS_BEARER_TOKEN_BEDROCK — Long-term API keys from Bedrock console
• AWS_WEB_IDENTITY_TOKEN_FILE / AWS_ROLE_ARN — EKS IRSA / Kubernetes OIDC

Authentication precedence:

1. Bearer Token (AWS_BEARER_TOKEN_BEDROCK or /connect token)
2. AWS Credential Chain (profile, access keys, shared credentials, IAM roles, Web Identity, instance metadata)

Custom inference profiles:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "amazon-bedrock": {
      "models": {
        "anthropic-claude-sonnet-4.5": {
          "id": "arn:aws:bedrock:us-east-1:xxx:application-inference-profile/yyy"
        }
      }
    }
  }
}

Azure OpenAI

Requires Azure OpenAI resource and model deployment.

Setup:

1. Create Azure OpenAI resource in Azure portal
2. Deploy a model in Azure AI Foundry (deployment name must match model name)
3. Run /connect, search for Azure
4. Enter API key
5. Set resource name:

export AZURE_RESOURCE_NAME=XXX

1. Run /models to select deployed model

Note: If you see "I'm sorry, but I cannot assist with that request" errors, change the content filter from DefaultV2 to Default in your Azure resource.

Azure Cognitive Services uses a separate provider with AZURE_COGNITIVE_SERVICES_RESOURCE_NAME.

GitLab Duo

Requires GitLab Premium or Ultimate subscription. Supports OAuth and personal access token.

Setup:

1. Run /connect, select GitLab
2. Choose authentication method:

• OAuth (Recommended): Browser opens for authorization
• Personal Access Token: Generate at GitLab User Settings > Access Tokens (scopes: api, token starts with glpat-)

1. Run /models to select model

Available models:

• duo-chat-haiku-4-5 (default) — Fast responses
• duo-chat-sonnet-4-5 — Balanced performance
• duo-chat-opus-4-5 — Most capable

Self-hosted GitLab:

export GITLAB_INSTANCE_URL=https://gitlab.company.com
export GITLAB_TOKEN=glpat-...
export GITLAB_AI_GATEWAY_URL=https://ai-gateway.company.com  # optional

Compliance config (lock to self-hosted):

{
  "$schema": "https://opencode.ai/config.json",
  "small_model": "gitlab/duo-chat-haiku-4-5",
  "share": "disabled"
}

OAuth for self-hosted: Create application with callback URL http://127.0.0.1:8080/callback and scopes api, read_user, read_repository. Set GITLAB_OAUTH_CLIENT_ID env var.

GitLab API tools plugin:

{
  "$schema": "https://opencode.ai/config.json",
  "plugin": ["opencode-gitlab-plugin"]
}

---

Cloud Providers

Cloudflare AI Gateway

Unified endpoint for OpenAI, Anthropic, Workers AI, and more.

Setup:

1. Create gateway in Cloudflare dashboard > AI > AI Gateway
2. Run /connect, search for Cloudflare AI Gateway
3. Enter Account ID, Gateway ID, and API token

Environment variables:

export CLOUDFLARE_ACCOUNT_ID=your-32-character-account-id
export CLOUDFLARE_GATEWAY_ID=your-gateway-id
export CLOUDFLARE_API_TOKEN=your-api-token

Config:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "cloudflare-ai-gateway": {
      "models": {
        "openai/gpt-4o": {},
        "anthropic/claude-sonnet-4": {}
      }
    }
  }
}

Cloudflare Workers AI

Run AI models on Cloudflare's global network via REST API.

Environment variables:

export CLOUDFLARE_ACCOUNT_ID=your-32-character-account-id
export CLOUDFLARE_API_KEY=your-api-token

DigitalOcean

Supports OAuth (recommended) and Model Access Keys. Inference Routers route requests to optimal models.

OAuth (recommended):

1. Run /connect, select DigitalOcean, choose Login with DigitalOcean
2. Authorize in browser
3. Inference Routers appear as router: in model picker

Model Access Key:

export DIGITALOCEAN_ACCESS_TOKEN=your-model-access-key

Note: Inference Routers are only auto-discovered with OAuth, not with Model Access Keys.

Vercel AI Gateway

Unified endpoint for OpenAI, Anthropic, Google, xAI, and more. Models at list price.

Setup:

1. Create API key in Vercel dashboard > AI Gateway tab
2. Run /connect, search for Vercel AI Gateway
3. Enter API key

Routing options:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "vercel": {
      "models": {
        "anthropic/claude-sonnet-4": {
          "options": {
            "order": ["anthropic", "vertex"],
            "only": ["anthropic"],
            "zeroDataRetention": false
          }
        }
      }
    }
  }
}

SAP AI Core

Access 40+ models from OpenAI, Anthropic, Google, Amazon, Meta, Mistral, and AI21.

Setup:

1. Create service key in SAP BTP Cockpit
2. Run /connect, search for SAP AI Core
3. Enter service key JSON

Environment variables:

export AICORE_SERVICE_KEY='{"clientid":"...","clientsecret":"...","url":"...","serviceurls":{"AI_API_URL":"..."}}'
export AICORE_DEPLOYMENT_ID=your-deployment-id  # optional
export AICORE_RESOURCE_GROUP=your-resource-group  # optional

---

AI Platform Providers

OpenRouter

Multi-provider gateway with many preloaded models.

Setup:

1. Create API key at openrouter.ai/settings/keys
2. Run /connect, search for OpenRouter
3. Enter API key

Config — add models and provider routing:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "openrouter": {
      "models": {
        "moonshotai/kimi-k2": {
          "options": {
            "provider": {
              "order": ["baseten"],
              "allow_fallbacks": false
            }
          }
        }
      }
    }
  }
}

Together AI

1. Create account at api.together.ai, click Add Key
2. Run /connect, search for Together AI, enter key
3. Run /models to select

Fireworks AI

1. Create account at app.fireworks.ai, click Create API Key
2. Run /connect, search for Fireworks AI, enter key
3. Run /models to select

Groq

1. Create account at console.groq.com, click Create API Key
2. Run /connect, search for Groq, enter key
3. Run /models to select

Cerebras

1. Create account at inference.cerebras.ai, generate API key
2. Run /connect, search for Cerebras, enter key
3. Run /models to select (e.g. Qwen 3 Coder 480B)

DeepSeek

1. Create account at platform.deepseek.com, create API key
2. Run /connect, search for DeepSeek, enter key
3. Run /models to select (e.g. DeepSeek V4 Pro)

Hugging Face

Inference Providers access via 17+ providers.

1. Create token at huggingface.co/settings/tokens with Inference Providers permission
2. Run /connect, search for Hugging Face, enter token
3. Run /models to select

NVIDIA

Free access to Nemotron and open models via build.nvidia.com.

1. Create account at build.nvidia.com, generate API key
2. Run /connect, search for NVIDIA, enter key
3. Run /models to select

Environment variable:

export NVIDIA_API_KEY=nvapi-your-key-here

On-prem / NIM:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "nvidia": {
      "options": {
        "baseURL": "http://localhost:8000/v1"
      }
    }
  }
}

Nebius Token Factory

1. Create account at tokenfactory.nebius.com, click Add Key
2. Run /connect, search for Nebius Token Factory, enter key
3. Run /models to select

Baseten

1. Create account at app.baseten.co, generate API key
2. Run /connect, search for Baseten, enter key
3. Run /models to select

Deep Infra

1. Create account at deepinfra.com/dash, generate API key
2. Run /connect, search for Deep Infra, enter key
3. Run /models

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Timmy6942025
• Source: Timmy6942025/opencode-builder-skill
• License: MIT

Install and usage instructions live in the source repository linked above.