Whizi for developers and agents

Whizi developer resources.

Every machine-readable surface Whizi publishes: the OpenAPI spec, versioned model cost data with a deprecation policy, llms.txt for AI agents, and rate limit conventions.

The Whizi OpenAPI spec

The spec at /openapi.json describes every public data endpoint with typed schemas, operation ids, the versioning and deprecation policy, and the error model. Whizi has no public product API: the endpoints below are unauthenticated static data with CORS enabled, and anything under /api/ on this host answers with an RFC 9457 problem+json 404.

Whizi data endpoints

Integrate against the versioned paths: /data/v1/model-cost-index.json and /data/v1/model-cost-index.csv. The dataset prices one standard answer (1,000 input tokens plus 500 output tokens) across 100 models from 29 providers, refreshed with each deploy, and is free to reuse under CC BY 4.0 with attribution. Within v1, fields are only added, never renamed or removed; a breaking change would ship as /data/v2/ with 6 months of Deprecation and Sunset notice on v1.

Whizi for AI agents: llms.txt

/llms.txt is a curated Markdown index of the whole site: product, pricing, docs, comparisons, and data. /llms-full.txt expands it with the direct answer and key facts from every page. Nonexistent paths return a real HTTP 404 that points back to these files, as JSON under /api/ and as Markdown when requested with Accept: text/markdown.

Rate limits

These are static files on a CDN with no per-client quota enforcement. The declared advisory policy on data responses is RateLimit-Policy: 1000;w=3600. Stay under 1,000 requests per hour per client and no throttling will ever apply; no 429s are currently issued.

Questions

Email [email protected]. For what Whizi is and who builds it, see About Whizi. Product documentation for humans starts at /docs/.