Use OpenMayhem from your code or your AI app

1. 1Create a keyIn your dashboard, create an API key, choose what it may be used for and set spending limits. The key is shown only once.Dashboard, API keys
2. 2Pick a modelFind a model in Markets and copy its ID. The market page shows the price and who sells it.Markets
3. 3Send a requestCall the API from your code, or ask your AI app once MCP is set up.The examples are below.

curl https://api.openmayhem.ai/v1/chat/completions \
  -H "Authorization: Bearer $OPENMAYHEM_API_KEY" \
  -H "Content-Type: application/json" \
  -H "X-OpenMayhem-Max-Cost-Usd: 0.02" \
  -d '{
    "model": "qwen/qwen3.8-flash-next",
    "messages": [{ "role": "user", "content": "Three names for a bakery by the sea" }]
  }'

MCP setup for every app

Choosing who serves a request

If you send none of these, the network chooses a seller for you. In an AI app you can ask in plain words, for example "use only T4 sellers and spend at most 2 cents".

The MCP arguments sit inside the routing object of a paid tool. max_cost_usd is a separate argument, and every paid tool requires it.

A proxied offer ID has three 64-character parts. The request also needs the proxy object that a saved routing profile returns when you resolve it. Proxied offers never receive native requests, and native sellers never receive proxied ones. Proxied requests accept the cost cap, minimum context, minimum speed and seller list from this table.

Routing in detailBuying from proxied sellers

Endpoints and MCP tools

Each entry shows the API endpoint and the MCP tool for the same task.

• Chat

  APIPOST /v1/chat/completionsMCPchat

  Text replies with tools, JSON output and streaming.

• Responses

  APIPOST /v1/responsesMCPcreate_response

  The same models in the OpenAI Responses format.

• Decisions

  APIPOST /v1/decisionsMCPdecide

  Typed answers such as yes or no, a choice or a score.

• Embeddings

  APIPOST /v1/embeddingsMCPembed

  Vectors for search and memory.

• Images

  APIPOST /v1/images/generationsMCPgenerate_image

  Images from text. These run as jobs.

• Video

  APIPOST /v1/videosMCPgenerate_video

  Clips from text or from a start image. These run as jobs.

• Speech

  APIPOST /v1/audio/speechMCPgenerate_speech

  Spoken audio from text.

• Transcription

  APIPOST /v1/audio/transcriptionsMCPtranscribe_audio

  Text from recorded speech, with timestamps.

• Music

  APIPOST /v1/music/generationsMCPgenerate_music

  Songs and instrumentals.

• Workflows

  APIPOST /v1/workflowsMCPrun_workflow

  Approved ComfyUI workflows for images, video and audio.

• Models and prices

  APIGET /v1/modelsMCPlist_models · get_model

  Every model with its price, sellers and status.

• Jobs and results

  APIGET /v1/jobs/{id}MCPget_job · cancel_job · get_artifact

  Check media jobs, cancel them and download the results.