# Lightning Rod Labs - Predict anything with AI

We train LLMs on messy, real-world data. Our small models out-predict frontier models at a fraction of the cost. Try our public Foresight Models today, or train an unmatched AI expert for your domain - using the messy data you already have.

Our latest forecasting model Foresight v4 outperforms all benchmarked SOTA models on price & accuracy on a benchmark of general prediction questions from Polymarket.

## Key Resources

For agents:
- Base API URL: `https://api.lightningrod.ai/v1`
- OpenAPI spec: https://api.lightningrod.ai/openapi.json
- Docs: https://docs.lightningrod.ai/llms.txt

For humans:
- Human dashboard: https://dashboard.lightningrod.ai
- Foresight models (inference API): https://www.lightningrod.ai/models
- Custom models (enterprise platform): https://www.lightningrod.ai/custom-models

To learn more about how we can work with you to build and deploy custom prediction models for your business, book a call with us: https://calendly.com/d/ctq4-7gd-nyq/lightning-rod-demo

## Agentic access (pay with MPP)

Agents obtain an API key and credits without a human, then call the standard API:

1. `POST /v1/mpp/topup` with an MPP payment → returns `{ organization_id, api_key, credited_cents }`. With no payment it returns `402` + a `WWW-Authenticate: Payment` challenge (fixed $5 USD top-up via a Stripe Shared Payment Token).
2. Use the returned `api_key` as `Authorization: Bearer` on `POST /openai/chat/completions`.
3. When the balance is exhausted the API returns `402`; call `/mpp/topup` again (send your key in `X-API-Key` to refill the same account).