Providers and models in Pi

Pi supports 20+ LLM providers. For each built-in provider, Pi maintains a curated list of tool-capable models, updated with every release. Authenticate via an OAuth subscription (/login) or an API key, then select any model from that provider via /model or Ctrl+L.

Subscription providers

Three providers support OAuth-based subscription login. Run /login inside Pi, then select a provider:

/login

Provider	Requirement
Claude Pro/Max	Anthropic Claude Pro or Max subscription
ChatGPT Plus/Pro (Codex)	OpenAI ChatGPT Plus or Pro subscription
GitHub Copilot	Active GitHub Copilot subscription

Anthropic subscription auth draws from extra usage and is billed per token, not against your Claude plan limits.

Use /logout to clear credentials. Tokens are stored in ~/.pi/agent/auth.json and refresh automatically when expired.

API key providers

Environment variables

Set an API key before launching Pi, and Pi will detect it automatically:

export ANTHROPIC_API_KEY=sk-ant-...
pi

The full list of supported providers and their environment variables:

Provider	Environment Variable	`auth.json` key
Anthropic	`ANTHROPIC_API_KEY`	`anthropic`
OpenAI	`OPENAI_API_KEY`	`openai`
Azure OpenAI Responses	`AZURE_OPENAI_API_KEY`	`azure-openai-responses`
DeepSeek	`DEEPSEEK_API_KEY`	`deepseek`
Google Gemini	`GEMINI_API_KEY`	`google`
Mistral	`MISTRAL_API_KEY`	`mistral`
Groq	`GROQ_API_KEY`	`groq`
Cerebras	`CEREBRAS_API_KEY`	`cerebras`
Cloudflare AI Gateway	`CLOUDFLARE_API_KEY` (+ `CLOUDFLARE_ACCOUNT_ID`, `CLOUDFLARE_GATEWAY_ID`)	`cloudflare-ai-gateway`
Cloudflare Workers AI	`CLOUDFLARE_API_KEY` (+ `CLOUDFLARE_ACCOUNT_ID`)	`cloudflare-workers-ai`
xAI	`XAI_API_KEY`	`xai`
OpenRouter	`OPENROUTER_API_KEY`	`openrouter`
Vercel AI Gateway	`AI_GATEWAY_API_KEY`	`vercel-ai-gateway`
ZAI	`ZAI_API_KEY`	`zai`
OpenCode Zen	`OPENCODE_API_KEY`	`opencode`
OpenCode Go	`OPENCODE_API_KEY`	`opencode-go`
Hugging Face	`HF_TOKEN`	`huggingface`
Fireworks	`FIREWORKS_API_KEY`	`fireworks`
Kimi For Coding	`KIMI_API_KEY`	`kimi-coding`
MiniMax	`MINIMAX_API_KEY`	`minimax`
MiniMax (China)	`MINIMAX_CN_API_KEY`	`minimax-cn`
Xiaomi MiMo	`XIAOMI_API_KEY`	`xiaomi`
Xiaomi MiMo Token Plan (China)	`XIAOMI_TOKEN_PLAN_CN_API_KEY`	`xiaomi-token-plan-cn`
Xiaomi MiMo Token Plan (Amsterdam)	`XIAOMI_TOKEN_PLAN_AMS_API_KEY`	`xiaomi-token-plan-ams`
Xiaomi MiMo Token Plan (Singapore)	`XIAOMI_TOKEN_PLAN_SGP_API_KEY`	`xiaomi-token-plan-sgp`

Auth file

You can also store credentials permanently in ~/.pi/agent/auth.json. The file is created with 0600 permissions (user read/write only). Credentials in auth.json take priority over environment variables.

{
  "anthropic": { "type": "api_key", "key": "sk-ant-..." },
  "openai": { "type": "api_key", "key": "sk-..." },
  "deepseek": { "type": "api_key", "key": "sk-..." },
  "google": { "type": "api_key", "key": "..." },
  "opencode": { "type": "api_key", "key": "..." }
}

You can also use /login in interactive mode and select an API key provider to store the key in auth.json without editing the file manually.

Key resolution

The key field in auth.json supports three formats:

Shell command

Prefix the value with ! to run a shell command and use its stdout. The result is cached for the process lifetime.

{ "type": "api_key", "key": "!security find-generic-password -ws 'anthropic'" }
{ "type": "api_key", "key": "!op read 'op://vault/item/credential'" }

Environment variable name

Provide the name of an environment variable (no $ prefix). Pi reads the variable at runtime.

{ "type": "api_key", "key": "MY_ANTHROPIC_KEY" }

Literal value

Any other string is used directly as the API key.

{ "type": "api_key", "key": "sk-ant-..." }

OAuth credentials are also stored in auth.json after /login and managed automatically by Pi.

Cloud providers

Azure OpenAI

export AZURE_OPENAI_API_KEY=...
export AZURE_OPENAI_BASE_URL=https://your-resource.openai.azure.com
# Or use the resource name instead of the full URL
export AZURE_OPENAI_RESOURCE_NAME=your-resource

# Optional
export AZURE_OPENAI_API_VERSION=2024-02-01
export AZURE_OPENAI_DEPLOYMENT_NAME_MAP=gpt-4=my-gpt4,gpt-4o=my-gpt4o

Both https://your-resource.openai.azure.com and https://your-resource.cognitiveservices.azure.com root endpoints are supported and auto-normalized to /openai/v1.

Amazon Bedrock

AWS profile
IAM keys
Bearer token

export AWS_PROFILE=your-profile

export AWS_ACCESS_KEY_ID=AKIA...
export AWS_SECRET_ACCESS_KEY=...

export AWS_BEARER_TOKEN_BEDROCK=...

Set the region (defaults to us-east-1):

export AWS_REGION=us-west-2

Pi also supports ECS task roles (AWS_CONTAINER_CREDENTIALS_*) and IRSA (AWS_WEB_IDENTITY_TOKEN_FILE). To use a specific Bedrock model:

pi --provider amazon-bedrock --model us.anthropic.claude-sonnet-4-20250514-v1:0

Prompt caching is enabled automatically for Claude models whose ID contains a recognizable model name. For application inference profiles (whose ARNs don’t contain the model name), force caching manually:

export AWS_BEDROCK_FORCE_CACHE=1
pi --provider amazon-bedrock --model arn:aws:bedrock:us-east-1:123456789012:application-inference-profile/abc123

Google Vertex AI

Uses Application Default Credentials (ADC):

gcloud auth application-default login
export GOOGLE_CLOUD_PROJECT=your-project
export GOOGLE_CLOUD_LOCATION=us-central1

Or set GOOGLE_APPLICATION_CREDENTIALS to a service account key file path.

Custom providers

Via models.json

Add custom providers and local models (Ollama, LM Studio, vLLM, or any OpenAI/Anthropic/Google-compatible API) via ~/.pi/agent/models.json. No restart required — the file reloads each time you open /model.

{
  "providers": {
    "ollama": {
      "baseUrl": "http://localhost:11434/v1",
      "api": "openai-completions",
      "apiKey": "ollama",
      "models": [
        { "id": "llama3.1:8b" },
        { "id": "qwen2.5-coder:7b" }
      ]
    }
  }
}

Supported API types: openai-completions, openai-responses, anthropic-messages, google-generative-ai. You can also use models.json to override a built-in provider’s base URL (for proxies) or merge additional models into an existing provider.

Via extensions

For providers that need custom API implementations or OAuth flows (enterprise SSO, corporate proxies), create an extension using pi.registerProvider(). See extensions for how to build one.

Resolution order

When Pi resolves credentials for a provider, it checks in this order:

CLI --api-key flag
auth.json entry (API key or OAuth token)
Environment variable
Custom provider keys from models.json

Get Started

Using Pi

Customization

Programmatic Usage

Providers and models in Pi

Subscription providers

API key providers

Environment variables

Auth file

Key resolution

Cloud providers

Azure OpenAI

Amazon Bedrock

Google Vertex AI

Custom providers

Via models.json

Via extensions

Resolution order

Build docs developers (and LLMs) love

Get Started

Using Pi

Customization

Programmatic Usage

Documentation Index

​Subscription providers

​API key providers

​Environment variables

​Auth file

​Key resolution

​Cloud providers

​Azure OpenAI

​Amazon Bedrock

​Google Vertex AI

​Custom providers

​Via models.json

​Via extensions

​Resolution order

Build docs developers (and LLMs) love

Subscription providers

API key providers

Environment variables

Auth file

Key resolution

Cloud providers

Azure OpenAI

Amazon Bedrock

Google Vertex AI

Custom providers

Via models.json

Via extensions

Resolution order