Provider API

Autohand AI model access

Select the Autohand AI provider, sign in or configure an Autohand API key, then run /model. The picker loads models allowed by your account. The authenticated GET /v1/models endpoint returns the same plan-filtered catalog. Weka uses /v1/decisions and is not available in CLI chat.

Compare all seven plan prices and limits, including Lite and Pro+.

Calling the API

Create an API key from the API Keys page in the console. Each key is shown only once, so copy and store the key securely when it is created.

curl https://api.autohand.ai/v1/chat/completions \
  -H "Authorization: Bearer $AUTOHAND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "fantail",
    "messages": [{"role": "user", "content": "Explain this stack trace"}]
  }'

Use a model ID available to your plan. See Available models for the full access comparison.

Shared usage

Hosted models share your plan allowance across Desktop, CLI, web, mobile, editor integrations, SDK and API. Switching models does not create a new allowance. Team and Enterprise credit limits are independent per member, not pooled. BYOK and local model requests do not consume Autohand credits.

One credit counts one successful hosted model response. A request rejected before model work begins does not use a credit. Credit allowances, requests per minute, and input/output token rate limits are separate controls. Model context length is a model property, not a subscription allowance.

Bring your own provider

Autohand Code can also connect to an external provider, including OpenRouter, OpenAI, DeepSeek, Cerebras, Azure AI Foundry, Google Cloud Vertex AI, and local runtimes such as Ollama, MLX, and llama.cpp.

When you use your own provider, that provider bills you directly and Autohand credit limits do not apply to those requests. Your Autohand plan and limits continue to apply whenever you use Autohand-hosted models.