RentPrompts API
Changelog
What changed in the API, newest first. Also see the API status and the API reference.
Versioning and deprecation policy
- One stable version
- Every endpoint lives under /api/v1. Within v1 we only make additive changes: new endpoints, new optional request fields, new response fields and new error codes. Parse responses leniently and ignore fields you do not know.
- Breaking changes
- A change that could break a working integration — removing a field or endpoint, changing a field's type, or making an optional field required — ships only in a new version (/api/v2). v1 keeps working for at least 12 months after v2 is released.
- Deprecations
- Deprecated behaviour is announced here and keeps working for at least 90 days before removal. Requests that use it are listed in your usage so you can find them.
- Models
- Models are added and retired as providers release and retire them. A retired model id answers 404 model_not_found; GET /v1/models is always the current list. Prices follow the website and are shown in the catalog before they apply.
Changes
Live streaming, function calling, webhooks and key rotation
- addedReal token streaming on /v1/chat/completions for models with capabilities.streaming in GET /v1/models.
- addedtools, tool_choice, parallel_tool_calls, tool messages and response_format json_schema on models with capabilities.tools.
- addedstream_options.include_usage: a final chunk with usage and rentprompts.cost_joules.
- addedA webhook URL on /v1/run and /v1/images/generations: signed job.succeeded / job.failed events (Standard Webhooks) with retries for ~35 hours.
- addedcapabilities object on every model in GET /v1/models.
- addedEdit a key's name, model types and monthly cap; rotate a key in one step; per-key webhook signing secret.
- addedOpenAPI 3.1 spec at /apidocs/openapi.json, Markdown docs at /apidocs.md, and a status page.
- changedChat conversations on native models are sent with their roles and history instead of one combined prompt.
- changedAn Idempotency-Key on a streamed request only blocks concurrent duplicates; streams are not stored for replay.
Usage by key and by model maker
- addedUsage filters (24 hours, 7, 30 and 90 days, custom dates) and breakdowns by key, model and maker on the API keys page.
- addedinputs and outputs on every model in GET /v1/models.
- changedKey management moved from /apidocs to /dashboard/api-keys.
Developer API launch
- addedrp_live_ keys with model-type scopes, monthly spend caps and expiry.
- added/v1/chat/completions, /v1/images/generations, /v1/run, /v1/jobs/{id}, /v1/models, /v1/balance and /v1/usage.
- addedIdempotency-Key support, typed error codes, request ids and cost headers.
- deprecatedProfile API keys (UUID). They keep working on /api/v1/* only; create an rp_live_ key instead.