Like Cursor. Any provider you want.

Same editor layout and agent workflow — now with Orbit Provider (GitHub sign-in) or the keys and local models you already use.

Available on macOS & Windows

One editor. Every AI/LLM provider.

AnthropicOpenAIGoogleDeepSeekQwenMetaMistralMoonshotOllamaCustom endpointAnthropicOpenAIGoogleDeepSeekQwenMetaMistralMoonshotOllamaCustom endpoint

Orbit works with Anthropic, OpenAI, Google, DeepSeek, Qwen, Meta, Mistral, Moonshot, Ollama, and any custom endpoint.

The workflow

Ask. Plan. Agent.

The three modes you already know from Cursor, in one sidebar, running on any provider you connect.

01

Ask

Understand the code

Questions about your repo, answered from the files themselves. No copy-paste, no context-switching.

Where is session validation handled?

Auth runs through lib/session.ts. Middleware checks the cookie before protected routes.
@lib/session.ts
Add follow-up…
Ask
02

Plan

Review before it ships

Larger changes start as a plan you review and steer. Approve the approach, then let the agent build.

Improve indexing for large monorepos

What should the plan focus on first?

  1. 1. Cache parsed symbols per revision
  2. 2. Parallel workers for large repos
  3. 3. Both: cache first, then parallelize
SkipContinue
Add follow-up…
Plan
03

Agent

Edits across files

Multi-file edits, shell commands, and tool calls, with you in the loop at every step.

Wire up parallel index workers

  • Read indexWorker.ts
  • Grep "workerPool" in src/
  • Edit indexWorker.ts
  • Shell npm run test:indexer

indexWorker.ts +12 −3

Review changes before accepting.

Add follow-up…
Agent

Providers

Orbit Provider or your keys. Switch every chat.

Sign in with GitHub for managed models — or bring Anthropic, OpenAI, Google, open-source providers, and local endpoints. Switch per conversation.

  • Orbit Provider

    Sign in with GitHub. Managed models — including GLM-5.2 at 50% off — with wallet billing. No API key to paste.

  • API keys

    Bring keys from any provider. Stored locally on your machine, never on our servers.

  • Local providers

    Run Llama, Qwen, or DeepSeek through Ollama or your own endpoint. No cloud required.

Works with Orbit Provider, Anthropic, OpenAI, Google, DeepSeek, Qwen, Meta, Mistral, and Ollama.

Why Orbit

The editor you expect. Full control underneath.

The AI code editor experience you already know, with you in charge of which provider powers it, and where your data goes.

Zero learning curve

Same layout, chat sidebar, and agent modes as Cursor. Open a project and start coding. Nothing new to learn.

Orbit Provider or your keys

Sign in with GitHub for managed models, or connect Anthropic, OpenAI, Google, OpenRouter, and more — switch per chat.

Open-source ready

Run Llama, Qwen, or DeepSeek locally via Ollama. Full agent workflow, no cloud required.

Star on GitHub

Your keys, your control

BYO API keys stay on your machine. Orbit Provider traffic uses your signed-in account — you choose the path.

FAQ

Quick answers

Everything you might want to know before you download. Still stuck? Ask on GitHub.

Try Orbit 0.4.0 on macOS or Windows

Sign in with GitHub for Orbit Provider — GLM-5.2 at 50% off — or bring your own keys. Same Cursor-like workflow either way.

Available on macOS & Windows