Skip to content

OpenAI

Autobot supports OpenAI as an LLM provider via the Chat Completions API. This gives access to the GPT-5 family, o3, and other OpenAI models. The mini and nano variants offer a great balance of capability and cost for everyday tasks.

Setup

1. Get an API key

Create an API key at platform.openai.com/api-keys.

2. Configure credentials

Add your API key to the .env file:

OPENAI_API_KEY=sk-...

Or use the interactive setup:

autobot setup
# Select "OpenAI (GPT)" as provider

3. Configure the provider

In config.yml:

agents:
  defaults:
    model: "openai/gpt-5-mini"

providers:
  openai:
    api_key: "${OPENAI_API_KEY}"

4. Verify

autobot doctor
# Should show: ✓ LLM provider configured (openai)

Model naming

Models use the openai/ prefix followed by the OpenAI model ID:

# GPT-5 family (recommended)
model: "openai/gpt-5-mini"       # Fast, cheap, capable (recommended default)
model: "openai/gpt-5-nano"       # Fastest, cheapest — great for simple tasks
model: "openai/gpt-5.2"          # Flagship — smartest, most precise

# Reasoning
model: "openai/o3"               # Complex reasoning tasks

# Coding
model: "openai/gpt-5.3-codex"   # Best for agentic coding

# Legacy (still available in the API)
model: "openai/gpt-4.1"
model: "openai/gpt-4.1-mini"
model: "openai/gpt-4.1-nano"

gpt-5-mini is a good default for most use cases — significantly cheaper and faster than the flagship while still handling tool use, coding, and reasoning well. Use gpt-5-nano for high-volume, simpler tasks like summarization or classification.

The openai/ prefix tells autobot to route to the OpenAI API. It is stripped before sending to the API.

See the full model list in the OpenAI docs.

Custom API base

To use a self-hosted proxy, Azure OpenAI, or any OpenAI-compatible endpoint:

providers:
  openai:
    api_key: "${OPENAI_API_KEY}"
    api_base: "https://your-proxy.example.com/v1/chat/completions"

Extra headers

Pass additional headers with every request:

providers:
  openai:
    api_key: "${OPENAI_API_KEY}"
    extra_headers:
      X-Custom-Header: "value"

Configuration reference

Field Required Default Description
api_key Yes — OpenAI API key (sk-...)
api_base No https://api.openai.com/v1/chat/completions Custom API endpoint
extra_headers No — Additional HTTP headers for every request

How it works

OpenAI uses the standard Chat Completions API format, which is the baseline for most LLM providers:

  • Authorization: Bearer header for authentication
  • Standard message format with role and content fields
  • Function calling via tools array with tool_choice

Autobot detects OpenAI models by keywords (openai, gpt, o1, o3, o4) and routes to the correct endpoint automatically. All GPT-5 family models, o3, and legacy GPT-4.1 models are supported. Tools, MCP servers, plugins, and all other features work seamlessly.

Voice transcription

OpenAI provides the Whisper API for voice transcription. When OpenAI is configured, voice messages are automatically transcribed using whisper-1. No extra configuration is needed — the API key is reused from the provider config.

If both Groq and OpenAI are configured, Groq is preferred for transcription (faster, free tier available).

Known limitations

  • No streaming — Responses are returned in full after the model finishes generating.
  • Tool choice is always auto — There is no configuration to force a specific tool or disable tool use per-request.

Troubleshooting

Enable debug logging to see request/response details:

LOG_LEVEL=DEBUG autobot agent -m "Hello"

Look for:

  • POST https://api.openai.com/v1/chat/completions model=... — confirms provider is active
  • Response 200 (N bytes) — confirms API response
  • HTTP 4xx/5xx: ... — API errors with details

Common issues

"No LLM provider configured" — Check that api_key is set and non-empty in config.yml.

"API error: Incorrect API key provided" — Invalid or revoked API key. Verify at platform.openai.com/api-keys.

"API error: Rate limit reached" — Too many requests or tokens. Check your usage at platform.openai.com/usage.

"API error: The model does not exist" — Model ID is wrong, or you don't have access to the model. Check available models in your account.