Gateway · Studio · Workspace

Our next era of sharp inference.

Talon is a fast, self-hosted LLM platform. Chat in the browser with web search and file uploads, build in a workspace that puts chat beside a code editor, or point any OpenAI client at /v1.

Open Studio See how it works

Protocol: OpenAI
Streaming: SSE
Prompt logs: None

One engine, three ways in

Use Talon from a browser tab, a code editor, or your own app.

Studio

A chat app in your browser. Search the web, attach files, switch between text-only and multimodal, and keep every chat in a sidebar.

Open Studio →

Workspace

A project folder with a file explorer and code editor, next to the chat. Select code and ask Talon about it in place.

See it in action →

Gateway API

Drop-in /v1/chat/completions with streaming and bearer keys. Any OpenAI SDK works after a base URL swap.

Read the API →

What it does, step by step

A quick walkthrough of what you see and what happens at each step.

Step 01

Start a chat

Open Studio and type a prompt. No setup, no API key required to get a response locally — Talon routes it straight to the model.

Summarize this PDF for me

Sure — on it

Step 02

Give it context

Attach files or images, or let it search the web. Previews render inline for PDFs, Office docs and code, so you see exactly what Talon sees.

report-q3.pdf

Searching the web…

Step 03

Build in a Workspace

Spin up a project folder with a file explorer and editor beside the chat. Select code, ask about it in place, and Talon answers with full context.

src/

main.rs

lib.rs

utils.rs

fn run() {

// ask Talon →

}

Step 04

Ship it through the API

Point any OpenAI SDK at /v1. Swap the base URL, add a bearer key, and stream responses with "stream": true.

curl $BASE/chat/completions \
  -H "Authorization: Bearer KEY" \
  -d '{"model":"talon",
       "stream": true}'

Studio

Chat that gets things done

✓Web search with DuckDuckGo, Wikipedia and GitHub results folded into answers.
✓Upload files and images. Previews for PDF, Office documents and code.
✓Chat limits that shrink as a conversation grows, so you always know where you stand.
✓Download any chat at any time. Light, dark and coder themes.

Open Studio

Workspace

Build with the model beside you

✓A project explorer and Monaco editor with tabs, save and rename.
✓Select code and ask about it from a floating prompt, right where you are.
✓Chat docks on the right or floats free. Hide it when you want focus.
✓Each workspace is a chat with its own files. Start one from Studio's New Workspace.

New Workspace

See the real thing

Talon Studio chat with a sidebar of chats, a chat limit pill and a streamed reply — Studio: chats in a sidebar, a chat limit that shrinks as you go, and a model badge for the active extension.

Talon Workspace with a file explorer, a Markdown editor and the chat docked on the right — Workspace: the file explorer and editor sit beside the chat, with Save and Download in the top bar.

Talon Studio welcome screen with New Chat and New Workspace buttons — Start a New Chat or a New Workspace from the welcome screen.

Ship in one request

Use the model ID talon with your existing OpenAI SDK. Just swap the base URL.

✓ Base URL
✓ Header Authorization: Bearer KEY
✓ Streaming with "stream": true

Ready to strike?

Open Studio to chat or build a workspace, or grab an API key and start sending requests.

Open Studio Get an API key