Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/pewdiepie-archdaemon/odysseus/llms.txt

Use this file to discover all available pages before exploring further.

Odysseus Chat is the primary conversation interface, letting you talk to any OpenAI-compatible language model — whether it’s running on your own GPU with vLLM, served locally by Ollama or llama.cpp, or accessed via a cloud API like OpenAI, OpenRouter, or GitHub Copilot. All it takes is a base URL and, where required, an API key. Streaming responses, session history, file uploads, saved presets, and a per-chat model switcher are all included out of the box.

​
Supported providers

ProviderEndpoint format
vLLMhttp://localhost:8000/v1
llama.cpphttp://localhost:8080/v1
Ollamahttp://localhost:11434/v1
OpenRouterhttps://openrouter.ai/api/v1
OpenAIhttps://api.openai.com/v1
GitHub Copilothttps://api.githubcopilot.com
Any endpoint that speaks the OpenAI Chat Completions shape (including LM Studio, Llamafile, and Tabby ML) works the same way — add the base URL and Odysseus discovers available models automatically.

​
Features

​
Streaming responses

Responses stream token-by-token as they are generated. Token-per-second, input/output token counts, and context-window utilization are shown at the bottom of each message once the stream finishes.

​
Session history

Each conversation is saved as a named session. Past sessions are listed in the sidebar, sorted most-recent first, and are fully searchable. You can fork, rename, archive, and truncate sessions from the session menu.

​
File uploads (vision and PDF)

You can attach images and PDF documents directly to a chat message. Odysseus sends image files to the model as vision content and extracts text from PDFs for context. The default upload limit is 10 MB per attachment. The limit is configurable via the environment variable: 
ODYSSEUS_CHAT_UPLOAD_MAX_BYTES=20971520  # 20 MB
Supported file types for chat uploads:
  • Images: JPEG, PNG, GIF, WebP (vision-capable models only)
  • Documents: PDF (text extracted and sent as context)

​
Presets

Presets are saved combinations of a system prompt and model configuration. Access them from the toolbar above the chat input — click a preset to load it into the current session. Presets let you switch between “Creative Writer”, “Code Review”, “Concise Assistant”, or any custom persona without re-typing a system prompt each time. Create and manage presets in Settings → Presets.

​
Model switcher

Each session has an independent model selection. Switch models mid-session from the model picker in the chat toolbar — subsequent messages in the same session will use the newly selected model.

​
Adding a provider

1

Open Settings

Click the gear icon or navigate to Settings → Models.
2

Add endpoint

Enter the base URL of your provider (for example, http://localhost:11434/v1 for Ollama). If the provider requires authentication, paste the API key in the API Key field.
3

Select a model

Odysseus queries the endpoint for its model list. Pick a default model from the dropdown, then close Settings. The model is now available in the chat model picker.

​
Compare mode

The Compare tab lets you send the same prompt to two different models side-by-side, with optional blind mode so you can judge responses without knowing which model produced which. See the Compare tab in the main navigation — it pairs naturally with Chat sessions.
If you run Ollama on the host machine and Odysseus inside Docker, use the special Docker hostname instead of localhost:
http://host.docker.internal:11434/v1
Ollama must also be bound to all interfaces for Docker to reach it:
OLLAMA_HOST=0.0.0.0:11434 ollama serve

Build docs developers (and LLMs) love

Get started for freeTalk to us