Platform
One API for models, hosted products, and billing.
Caedral is an OpenAI-compatible AI API. One key, subscription pools, Caedral-operated Embed / Rerank / Voice, and Notre Runtime on eligible chat when the platform runs it.
QuickstartProduct architecture
Your application talks to Caedral. Caedral talks to models. You do not assemble provider accounts to ship.
Your application
↓
Caedral
- Introduction
What the API is: one key, two pools, HTTP 402, Notre on eligible chat.
- Quickstart
Create a key, pick a plan, send curl to POST /v1/chat/completions.
- API reference
OpenAI-compatible /v1 routes. Auth, errors, streaming.
- Notre Runtime
Included on every plan. Applied on eligible chat when the platform runs Notre.
- Caedral Embed
Hosted embeddings on the Caedral Models pool. Not Notre.
- Caedral Voice
English WAV speech on the Caedral Models pool. Not Notre.
- Billing
Two pools, on-demand in $5 increments, HTTP 402 at the hard stop.
↓AI workload
Built by Caedral
Not everything on Caedral comes from another provider.
- Caedral Embed
384-d embeddings for search and RAG. Caedral Models pool.
Available - Caedral Rerank
Hosted reranker after retrieval. External Models pool.
Available - Caedral Voice
English WAV speech. Caedral Models pool. No Notre.
Available - Notre Runtime
Native context compiler for eligible chat. Included on every plan. Not Embed, Rerank, or Voice.
Experimental
Operating the platform
- Teams
- Per-seat pooled usage, owner billing, member visibility. Teams docs
- Dashboard
- Keys, usage, logs, Stripe portal, bots, sellers. Create an account
- Notre
- Optional. Chat only. Notre program
- Status
- Probe-based availability for components and models. /status