Model choice is a setting, not a subscription.
A picker grouped by provider: OpenAI GPT-5 Codex, o4 mini, gpt-4o, gpt-3.5-turbo, Anthropic, and DeepSeek reasoner.
Supported models are marked Streaming, so long answers arrive token by token instead of all at once.
Group related conversations into collections so a project keeps its own thread history.
Attach files directly in the composer instead of pasting content in as text.
Code & Analysis, Learn & Explore, Write & Create, Plan & Organise. A first prompt for each mode of work.
A dense, low-glare workspace built for long sessions, with a light theme one click away.
Switch provider without switching tool.
The picker sits at the top of every chat, grouped by provider, with streaming support shown per model. Start a reasoning task on DeepSeek, run the code work on a Codex model, and keep both in the same sidebar.
Bring your own keys. The interface is free.
The workspace costs nothing. You connect your own provider keys, so model usage is billed by the provider, not by us.