# Astorias

> Astorias is an OpenAI-compatible inference API. Serve, batch, and fine-tune models with per-token pricing and zero data retention.

Astorias is the product. Saturated Labs Private Limited operates it. The inference API is OpenAI-compatible at https://api.astorias.ai/v1.

## Product

- [Home](https://astorias.ai/): Inference you can depend on. Serve, batch, and fine-tune.
- [Pricing](https://astorias.ai/pricing): Pay for tokens you use. No platform fee.
- [Models](https://astorias.ai/models): Hosted model ids and price per million tokens.
- [Blog](https://astorias.ai/blog): Engineering notes on serving, batching, and fine-tuning.
- [Status](https://astorias.ai/status): Uptime and incidents.

## Documentation

- [Docs](https://docs.astorias.ai/docs): OpenAI-compatible chat inference
- [Docs for agents](https://docs.astorias.ai/llms.txt)
- [Full docs for agents](https://docs.astorias.ai/llms-full.txt)

## Policies

- [Privacy](https://astorias.ai/legal/privacy): Zero retention of LLM content.
- [Terms](https://astorias.ai/legal/terms): Terms of use.

## Blog

- [Qwen3.8-27B is now live, also the winning recipes for 5090s are now public](https://astorias.ai/blog/qwen-3-8-27b-5090-recipes): Engineering note

## Optional

- [Full product brief](https://astorias.ai/llms-full.txt): FAQ and product facts for agents