# Taalas

> Taalas builds "Hardcore Models" — AI models etched directly into custom silicon instead of executed
> as software on general-purpose GPUs ("The Model is The Computer"). Its HC1 technology demonstrator
> hard-wires Meta's Llama 3.1 8B into an 815mm2, 53-billion-transistor die on TSMC 6nm, and is exposed
> to developers as a hosted inference REST API at api.taalas.com with two OpenAPI 3.1.0 contracts: a
> Taalas-native root surface (/health, /models, /generate) and an OpenAI-compatible /v1 surface
> (/v1/models, /v1/completions, /v1/chat/completions).

Generated by the API Evangelist enrichment pipeline on 2026-08-02. Taalas does not publish its own
llms.txt (/llms.txt returns 404 on taalas.com, api.taalas.com and chatjimmy.ai), so this file is
generated from the harvested apis.yml and repository artifacts.

## Access

- Authentication: `Authorization: Bearer <api-key>` — an opaque API key. Only `GET /health` is
  reachable anonymously; every other path returns 401.
- API keys are issued through the request form at https://taalas.com/api-request-form/, which as of
  2026-08-02 states applications are closed due to demand.
- Because /v1 mirrors the OpenAI request contract, OpenAI client libraries work by setting the base
  URL to https://api.taalas.com/v1.

## APIs

- [Taalas API](https://api.taalas.com/docs): Taalas-native inference — health, model info, and
  /generate with streaming, tool calling, guided JSON/regex generation, logprobs and stop sequences.
  Base URL https://api.taalas.com
- [Taalas API v1 (OpenAI-compatible)](https://api.taalas.com/v1/docs): /v1/models, /v1/completions
  and /v1/chat/completions with OpenAI-shaped parameters, multimodal text/image message content and
  function-calling tools. Base URL https://api.taalas.com/v1

## Specs

- [Taalas API OpenAPI 3.1.0](https://api.taalas.com/openapi.json)
- [Taalas API v1 OpenAPI 3.1.0](https://api.taalas.com/v1/openapi.json)

## Operations

- `health_health_get` — GET /health — server status, inference queue depth, current LoRA adapter (anonymous)
- `models_models_get` — GET /models — model info
- `generate_generate_post` — POST /generate — Taalas-native text generation
- `models_models_get` — GET /v1/models — model info (v1)
- `completions_completions_post` — POST /v1/completions — OpenAI-compatible completion
- `chat_completions_chat_completions_post` — POST /v1/chat/completions — OpenAI-compatible chat completion

## Docs

- [API home](https://api.taalas.com/)
- [Root API docs (Redoc)](https://api.taalas.com/docs)
- [v1 API docs (Redoc)](https://api.taalas.com/v1/docs)
- [Report a bug](https://api.taalas.com/bug-report)
- [Status](https://status.taalas.com/)

## Company

- [Website](https://taalas.com/)
- [Products — HC1](https://taalas.com/products/)
- [Mission Log](https://taalas.com/mission-log/)
- [The path to ubiquitous AI](https://taalas.com/the-path-to-ubiquitous-ai/)
- [Careers](https://taalas.com/position/)
- [ChatJimmy — public HC1 demo](https://chatjimmy.ai/)
- [Terms & Conditions](https://taalas.com/terms-conditions/)
- [Privacy Policy](https://taalas.com/privacy-policy/)
- Contact: info@taalas.com

## Optional

- [API Evangelist profile artifacts](https://github.com/api-evangelist/taalas) — authentication,
  conventions, error catalog, data model, lifecycle, conformance, agent skills and agentic-access
  contracts derived from the published specs.