The AI Tools API will expose Ansdev's curated language and audio models through a single billing-unified surface — drop one API key, debit the same wallet, no separate provider accounts to manage.
ℹ️ Under development
This API is not yet available. Capacity is being negotiated with upstream model providers. Email hello@ansdev.cloud to join the early access list.
Planned endpoints
| Method | Path | Purpose |
|---|---|---|
GET | /v1/ai/models | List available models with per-token pricing |
POST | /v1/ai/chat | Chat completions (streaming + non-streaming) |
POST | /v1/ai/transcribe | Audio → text transcription with optional diarisation |
POST | /v1/ai/embeddings | Vector embeddings for RAG pipelines |
Planned scopes
ai:compute— call any AI endpointai:read— list models, fetch usage stats only
Streaming
POST /v1/ai/chat will support server-sent events for streaming
generation:
Accept: text/event-stream
Mid-stream credit exhaustion will gracefully cut the stream at the
next chunk boundary with an explicit INSUFFICIENT_FUNDS event —
no silent truncation.
Billing model
Token-based pricing per model, debited from the AI Energy Core.
You set a hard daily cap in the dashboard; once hit, calls return
429 AI_QUOTA_EXCEEDED until midnight UTC.