Use case: GPT and assistant builders

Build a custom GPT with live sports odds

Wire a custom GPT Action to a live odds API with an OpenAPI schema import, key auth, and source-labeled answers.

Who this is for

Builders who want a ChatGPT assistant that answers with real current odds instead of guesses. The whole build is configuration: import a schema, set a header, and write instructions that make the GPT cite its sources.

Recommended workflow
  1. Create a custom GPT and add an Action
  2. Import the schema from /openapi.json
  3. Set API key auth with the X-API-Key header
  4. Instruct the GPT to name the book behind every price it quotes

The three endpoints this build uses

GET /openapi.jsonThe schema the Action imports. It describes every endpoint, so you can prune it to the handful the GPT should call.
GET /v1/sports/{sport_key}/oddsThe workhorse behind most user questions, with source-labeled prices the GPT should quote as is.
GET /v1/sports/{sport_key}/best-lineAnswers the most common assistant question, where is the best price, with the book named in the response.

Working code

# In the GPT editor: Actions > Import from URL
https://parlay-api.com/openapi.json

# Auth: API Key, header name X-API-Key, paste a free key from /signup.
# Instruction worth adding to the GPT:
# "Always state which bookmaker each quoted price comes from, and never
#  average prices across bookmakers."

Which tier fits

A free-tier key covers a personal GPT, and discovery endpoints cost nothing, so capability probing is free. A shared or public GPT burns credits with every user question, which is where paid tiers come in. Current limits are on the pricing page.

Build links

/answers/chatgpt-sports-odds/answers/odds-api-for-ai-agents/mcp