Branch any AI answer into focused sub-conversations — as deep as the idea goes. Chats, keys, and models live in your account.
Keys are stored in your account and used only by the server to call providers. They never return to your browser and are never called from it.
Pick which models fill your model selectors. Fetch the live catalog from any provider you have a key for — free models are tagged where the provider publishes pricing (OpenRouter does) — or add a model ID by hand.
Open Settings, paste an API key from OpenRouter, OpenAI, Anthropic, Google, xAI (Grok), Groq, Mistral, or DeepSeek, then fetch that provider's catalog. Use the filter and the Free only toggle to find no-cost models, and activate the ones you want — active models fill every model selector.
Enter sends, Shift+Enter adds a line, Esc or Stop cancels a streaming reply. Each chat remembers its model; everything autosaves to your account.
Select any text in an AI reply and click Branch. A focused sub-conversation opens with full parent context, and the selected text stays highlighted — click it any time to reopen. Threads nest without limit, each layer can use a different model, and every layer has its own color: the deeper you go, the hotter the hue.
Ctrl+K searches everything — scope it to chats, threads, or only deep branches. The Explorer (tree icon) maps every branch of one chat or all of them, with a filter for titles and quoted text. Breadcrumbs at the top of every thread trace the path back to the root.