Provider YAML Recipes

Drop one of these model: blocks into an agent or defaults.model. All string values support ${ENV_VAR} expansion.

:::tip Runnable provider files Standalone configurations live under examples/yaml-configs/providers/.

export ANTHROPIC_API_KEY=sk-ant-...
chronos -c examples/yaml-configs/providers/anthropic.yaml run "Hello"

:::

Provider matrix

Provider	provider value	Authentication
OpenAI	openai	OPENAI_API_KEY
Anthropic	anthropic	ANTHROPIC_API_KEY
Gemini AI Studio	gemini	GEMINI_API_KEY
Vertex AI	compatible	GOOGLE_ACCESS_TOKEN
Azure OpenAI	azure	AZURE_OPENAI_API_KEY
Grok / xAI	compatible	XAI_API_KEY
Groq	groq	GROQ_API_KEY
Mistral	mistral	MISTRAL_API_KEY
DeepSeek	deepseek	DEEPSEEK_API_KEY
Ollama	ollama	None
Other OpenAI-compatible APIs	compatible	Provider-specific key

OpenAI

model:
  provider: openai
  model: gpt-5.5
  api_key: ${OPENAI_API_KEY}
  timeout_sec: 120

Native reasoning:

reasoning:
  native: true
  effort: high
  summary: false

Anthropic

model:
  provider: anthropic
  model: claude-opus-4-8
  api_key: ${ANTHROPIC_API_KEY}
  timeout_sec: 120

Extended thinking for a tool-free agent:

reasoning:
  native: true
  budget_tokens: 4096
  summary: true

Anthropic native reasoning with tools currently fails closed because signed thinking blocks cannot yet be preserved across tool rounds. Use prompt strategy reasoning when tools are present.

Google Gemini (AI Studio)

model:
  provider: gemini
  model: gemini-2.0-flash
  api_key: ${GEMINI_API_KEY}

Thinking for a tool-free agent:

reasoning:
  native: true
  budget_tokens: 4096
  summary: true

Gemini native reasoning with tools has the same signed-thought limitation as Anthropic.

Google Vertex AI

Vertex exposes an OpenAI-compatible endpoint. Use a short-lived Google access token:

model:
  provider: compatible
  model: google/gemini-2.0-flash
  base_url: https://us-central1-aiplatform.googleapis.com/v1/projects/YOUR_PROJECT/locations/us-central1/endpoints/openapi
  api_key: ${GOOGLE_ACCESS_TOKEN}

export GOOGLE_ACCESS_TOKEN=$(gcloud auth print-access-token)

Azure OpenAI

Azure requires the resource endpoint, deployment name, and API version:

model:
  provider: azure
  deployment: my-gpt4o-deployment
  endpoint: https://my-resource.openai.azure.com
  api_version: "2024-10-21"
  api_key: ${AZURE_OPENAI_API_KEY}

Grok (xAI)

model:
  provider: compatible
  model: grok-4
  base_url: https://api.x.ai/v1
  api_key: ${XAI_API_KEY}

Groq

model:
  provider: groq
  model: llama-3.3-70b-versatile
  api_key: ${GROQ_API_KEY}

Mistral

model:
  provider: mistral
  model: mistral-large-latest
  api_key: ${MISTRAL_API_KEY}

DeepSeek

model:
  provider: deepseek
  model: deepseek-chat
  api_key: ${DEEPSEEK_API_KEY}

Ollama

model:
  provider: ollama
  model: llama3.2
  base_url: http://localhost:11434

ollama pull llama3.2
chronos -c examples/yaml-configs/providers/ollama.yaml run "Hello locally"

Other OpenAI-compatible endpoints

Use compatible for Together, OpenRouter, Fireworks, Perplexity, vLLM, LiteLLM, and other Chat Completions-compatible APIs:

model:
  provider: compatible
  model: meta-llama/Llama-3.3-70B-Instruct-Turbo
  base_url: https://api.together.xyz/v1
  api_key: ${TOGETHER_API_KEY}

Mix providers in one team

Each agent owns its own model block, so a parallel team can compare providers:

teams:
  - id: compare
    name: Provider Comparison
    strategy: parallel
    agents: [openai-agent, claude-agent, gemini-agent]
    max_concurrency: 3
    error_strategy: best_effort

See the complete multi-provider application.