Everything ChatGPT and Claude have, plus 100+ models on the same wallet — and a few things they still don't ship.
GPT-4o, Claude 3.5 Sonnet, Gemini 2.5, Llama 3.3, DeepSeek, Grok, Qwen — pick any model per message. Switch mid-conversation. One bill, not ten.
Auto-summarised facts about you carry between chats. Plus a custom-instructions field every model reads. Edit or wipe it any time.
Toggle the globe icon and your message goes out with live search results prepended. Works on every model — no native tool-use required.
Pick Gemini Flash Image or GPT-Image and prompt — get an image back. Drop in a photo and ask the model to edit it. Same wallet as text.
Drop in a PDF, paste a screenshot, or attach a code file. We extract the text server-side so even non-vision models can read it.
Dictate messages with the mic, hear replies read back. Conversation mode chains the loop hands-free. Works on every browser, including Firefox.
Group chats by topic. Project-level system prompt is prepended to every conversation in the folder so context carries automatically.
Save a prompt, pick a cadence (hourly / daily / weekly in your timezone), and we auto-fire it as a new chat when due — like ChatGPT Tasks.
When the model emits a full HTML page or SVG, click ▶ Preview to render it live in a side pane — Claude.ai-style, on every model.
Re-run the same turn with a different model in one click. Branch off any assistant message into a new conversation without losing the original.
Drop-in OpenAI-compatible POST /v1/chat/completions. Web search, image gen, every model — same token balance as the chat UI. See the docs →
Public share links, per-chat Markdown export, full account JSON dump in one click. Your data leaves with you.