# Astral Documentation

> Native inference runtime reference for applications and game engines

## Overview

- [Astral](https://astralruntime.dev/docs)

## Get Started

- [Building Astral](https://astralruntime.dev/docs/BUILD)
- [Feature matrix](https://astralruntime.dev/docs/docs/FEATURE_MATRIX)
- [Model paths](https://astralruntime.dev/docs/docs/api/MODEL_PATHS)
- [Model presets](https://astralruntime.dev/docs/docs/api/MODEL_PRESETS)
- [Error handling](https://astralruntime.dev/docs/docs/api/ERROR_HANDLING)

## Runtime APIs

- [Continuous batching](https://astralruntime.dev/docs/docs/api/CONTINUOUS_BATCHING)
- [Generation controls (C ABI)](https://astralruntime.dev/docs/docs/api/GENERATION_CONTROLS)
- [Tokenization API](https://astralruntime.dev/docs/docs/api/TOKENIZATION)
- [Prompt cache API](https://astralruntime.dev/docs/docs/api/PROMPT_CACHE)
- [Structured output and tools](https://astralruntime.dev/docs/docs/api/STRUCTURED_OUTPUT)
- [LoRA adapters](https://astralruntime.dev/docs/docs/api/LORA_ADAPTERS)
- [Native chunking](https://astralruntime.dev/docs/docs/api/CHUNKING)
- [Native memory index](https://astralruntime.dev/docs/docs/api/MEMORY_INDEX)
- [Native agents](https://astralruntime.dev/docs/docs/api/AGENT_RUNTIME)
- [Async delivery](https://astralruntime.dev/docs/docs/api/ASYNC_DELIVERY)
- [Remote runtime](https://astralruntime.dev/docs/docs/api/REMOTE_RUNTIME)
- [Hot-path profiling](https://astralruntime.dev/docs/docs/api/HOT_PATH_PROFILING)

## Architecture

- [Runtime architecture](https://astralruntime.dev/docs/docs/architecture/RUNTIME_ARCHITECTURE)
- [Retrieval architecture](https://astralruntime.dev/docs/docs/architecture/RETRIEVAL_ARCHITECTURE)
- [Backend architecture](https://astralruntime.dev/docs/docs/architecture/BACKEND_ARCHITECTURE)
- [Concurrency and threading model](https://astralruntime.dev/docs/docs/architecture/CONCURRENCY_MODEL)
- [Memory architecture](https://astralruntime.dev/docs/docs/architecture/MEMORY_ARCHITECTURE)
- [Low-level primitives](https://astralruntime.dev/docs/docs/architecture/LOW_LEVEL_PRIMITIVES)
- [Embedded / robotics profile](https://astralruntime.dev/docs/docs/EMBEDDED_PROFILE)
- [Embedded and robotics profile (Linux edge)](https://astralruntime.dev/docs/docs/embedded)
- [Vision + audio support](https://astralruntime.dev/docs/docs/VISION_AUDIO)

## Integration

- [Astral runtime Unity package](https://astralruntime.dev/docs/plugins/unity)
- [AstralRT (Unreal Engine plugin)](https://astralruntime.dev/docs/plugins/unreal/AstralRT)
- [AstralSample](https://astralruntime.dev/docs/examples/unreal/AstralSample)
- [Unreal 5.7 quickstart](https://astralruntime.dev/docs/docs/integration/UNREAL_57_QUICKSTART)

### Unity samples

- [Streaming chat](https://astralruntime.dev/docs/plugins/unity/Samples~/StreamingChat)
- [Multiple conversations](https://astralruntime.dev/docs/plugins/unity/Samples~/MultipleConversations)
- [Stateful NPC](https://astralruntime.dev/docs/plugins/unity/Samples~/StatefulNpc)
- [Local knowledge](https://astralruntime.dev/docs/plugins/unity/Samples~/LocalKnowledge)
- [Character variants](https://astralruntime.dev/docs/plugins/unity/Samples~/CharacterVariants)
- [Multimodal input](https://astralruntime.dev/docs/plugins/unity/Samples~/MultimodalInput)

## Profiling and CUDA

- [Tracy profiling](https://astralruntime.dev/docs/docs/PROFILING_TRACY)
- [CUDA parity](https://astralruntime.dev/docs/docs/CUDA_PARITY)
- [CUDA kernel strategy (ggml-cuda + cuBLAS)](https://astralruntime.dev/docs/docs/CUDA_KERNEL_STRATEGY)

## Release

- [ABI and versioning policy (draft)](https://astralruntime.dev/docs/docs/ABI_VERSIONING)
- [Release acceptance matrix](https://astralruntime.dev/docs/docs/release/RELEASE_ACCEPTANCE_MATRIX)
- [ABI manifest](https://astralruntime.dev/docs/docs/release/ABI_MANIFEST)
- [Dependency manifest](https://astralruntime.dev/docs/docs/release/DEPENDENCY_MANIFEST)
- [Third-Party notices](https://astralruntime.dev/docs/docs/release/THIRD_PARTY_NOTICES)
- [Astral <version> release notes](https://astralruntime.dev/docs/docs/release/RELEASE_NOTES_TEMPLATE)

## Project

- [Documentation](https://astralruntime.dev/docs/docs)
- [Contributing to Astral](https://astralruntime.dev/docs/CONTRIBUTING)
- [Security policy](https://astralruntime.dev/docs/SECURITY)
- [Changelog](https://astralruntime.dev/docs/CHANGELOG)
- [Astral roadmap](https://astralruntime.dev/docs/ROADMAP)
- [Coding standards](https://astralruntime.dev/docs/docs/rules/CODING_STANDARDS)

## Agent resources

- [llms-full.txt](https://astralruntime.dev/docs/llms-full.txt): The full Markdown of every page in one file.
- [Page Markdown](https://astralruntime.dev/docs/index.md): Append `.md` to any page URL to fetch that page as raw Markdown.
- [JSON API](https://astralruntime.dev/docs/api/docs/pages.json): Page index of the JSON docs API; each entry links the page's JSON and Markdown forms. Described by the OpenAPI document at https://astralruntime.dev/docs/openapi.json.
- [API catalog](https://astralruntime.dev/docs/.well-known/api-catalog): RFC 9727 linkset of the APIs documented here.
- [AI catalog](https://astralruntime.dev/docs/.well-known/ai-catalog.json): ARD manifest of the agent-facing resources on this site (MCP server, skills, APIs).
- [agent-readability.json](https://astralruntime.dev/docs/agent-readability.json): Manifest of every agent-facing artifact on this site.
- [Sitemap](https://astralruntime.dev/docs/sitemap.xml): Every indexable page URL with its last-modified date.