# NTS API documentation

[API Model AI catalog](/models)
[TTS / STT / Translation](/api-docs)
[Legacy Markdown](/api-docs/docs.md)
[Legacy Postman](/api-docs/postman.json)

API Model AI: read pricing, performance and API docs inside each published model. Use API Model keys from /api-model/keys.
Model Markdown: /api-docs/docs.md?model=<model-id>&lang=vi|en
Model Postman: /api-docs/postman.json?model=<model-id>&lang=vi|en

Public JSON catalog (no key, text/image/video, pagination): https://api.ntsgenai.com/v1/catalog/models
Limit: 60 requests/min/IP across pages and filters; respect Retry-After on 429. Prices are decimal strings. Reference uploads: POST /v1/media/uploads?type=image|video|audio (API Model key, multipart file).

## Published models
- [gemini-3.8-flash](/models?model=gemini-3.8-flash&tab=api) — [Markdown](/api-docs/docs.md?model=gemini-3.8-flash&lang=en)
- [gpt-image-2.5-sunburst](/models?model=gpt-image-2.5-sunburst&tab=api) — [Markdown](/api-docs/docs.md?model=gpt-image-2.5-sunburst&lang=en)
- [GPT Image 2](/models?model=gpt-image-2&tab=api) — [Markdown](/api-docs/docs.md?model=gpt-image-2&lang=en)

## TTS / STT / Translation
TTS/STT/Translation use the Account web_api_key via Authorization: Bearer. The two key types are not interchangeable. The old domain remains supported; switch to the new domain by changing only the hostname, preserving paths, keys and payloads. Existing plans, daily Credit, cache and billing apply: TTS per line character count, STT rounded up to minutes (minimum 1), translation per translated character. Completed portions may be charged even if a project fails or stops. API Model USD limits do not apply. Legacy creation POSTs have no Idempotency-Key guarantee; do not automatically repeat POST after losing a response.

New base: https://api.ntsgenai.com/api/v1/public-api
Old base: https://app.ntstool.com/api/v1/public-api