# Triton AI Docs

- [Triton AI documentation](/): Use Triton AI models, APIs, and campus services.

- **Choose a starting point**
- Services
  - [Services](/services): Choose a Triton AI service for chat, agent work, or repeatable workflows.
  - [TritonGPT](/services/tritongpt): Use UC San Diego's web workspace for chat, documents, and campus assistants.
  - [TritonAI Harness](/services/tritonai-harness): Use UC San Diego's managed desktop workspace for interactive agent work.
  - [n8n](/services/n8n): Build repeatable UC San Diego workflows with known triggers and steps.
- [Citizen developers](/citizen-developers): Plan a small campus AI project, choose a supported tool, and prepare it for review.

- **Build with models**
- Developer API
  - [Triton AI Developer API](/llm-api): Build with approved models through the Triton AI Developer API.

  - **Get started**
  - [Quickstart](/llm-api/quickstart): Send a chat request through the Triton AI Developer API.
  - [Models](/llm-api/models): Compare current cloud and UC San Diego-hosted models by task, limits, and price.
  - [Authentication](/docs/llm-api/api-reference#authenticate-a-request)
  - [Check usage](/usage)
  - Developer tools
    - [Integrations](/llm-api/integrations): Connect coding agents and developer tools to the Triton AI Developer API.
    - [Claude Code](/llm-api/integrations/claude-code): Connect Claude Code CLI, desktop, and VS Code to the Triton AI Developer API.
    - [Codex](/llm-api/integrations/codex): Connect Codex CLI, desktop, and VS Code to the Triton AI Developer API.
    - [Community integrations](/llm-api/integrations/community): Connect other developer tools to the Triton AI Developer API.

  - **Build with the API**
  - [Chat completions](/llm-api/chat-completions): Send text messages and stream responses with Python or cURL.
  - [Responses](/llm-api/responses): Create and stream responses with Python or cURL.
  - [Tool calling](/llm-api/tool-calling): Let a model request functions that your application controls.
  - [Image input](/llm-api/image-input): Send an image to a vision-capable model with Python or cURL.
  - [Embeddings](/llm-api/embeddings): Create text embeddings with the UC San Diego-hosted Qwen Embedding 4B route.

  - **Ship and troubleshoot**
  - [Production checklist](/llm-api/production): Prepare a Developer API integration for production use.
  - [Safety and security](/llm-api/safety): Classify requests against a written policy before model use.
  - [Errors and retries](/llm-api/errors): Handle Developer API failures, timeouts, and rate limits.

  - **Reference**
  - [API reference](/llm-api/api-reference): See the documented public routes for the Triton AI Developer API.
  - [Response headers](/docs/llm-api/api-reference#common-response-headers)

- **Account**
- [Check API usage](/usage)