Getting started

Enter your email on the sign-in page and open the link we send. It works for 15 minutes; there's no password. The first link creates your account on the Free plan: a one-time trial of 5 messages on every model but Claude Fable 5.1 and GPT-6 Astra (one of them can be on Claude Opus 5.5).

The model picker in the message box lists Claude, GPT, Gemini, DeepSeek, Grok, Kimi, GLM, and Mistral. The picker shows how many messages you have left on each model, and the message box shows it again before you send. The default is GPT-6 Luna (60 messages a day on Pro).

Switch models whenever you like: the next model sees the whole chat.

Attach images (JPEG, PNG, WebP, and GIF), PDFs, Word, Excel, and PowerPoint files, and text, data and code files (plain text, Markdown, and CSV, JSON, Python and more): up to 10 files a message, each up to 20 MB. Images can be up to 5 MB and PDFs up to 100 pages. Every model reads Word, Excel, and PowerPoint files as their text, each sheet of a workbook as a table. Models that don't read images skip them; models that don't take PDFs as files get the PDF's text.

• Documents

  Ask for a draft, code or a table and it opens beside the chat as a document that keeps every version. Text downloads as PDF, Word or Markdown; tables as Excel or CSV; code as its own file.

• HTML apps

  Ask for a calculator or a small tool and it runs live as an HTML app in the side panel.

• Web search and links

  The model can search the web and show its sources as cards, or read a link you paste. On Free, web search works inside your 5 trial messages; on a paid plan a search counts as one more message of the model in use.

• Personas and skills

  In Settings → Skills, save personas (instructions plus a default model; Writer, Coder and Tutor are built in) and upload skills the model can use.

• Memory

  Tell a model something lasting about you, or ask it to remember something, and it saves a short note (on models that can use tools); the reply shows “Memory updated” with Undo. Every model reads the same memory. Add, edit or delete memories in Settings → Memory, or turn memory off there; to keep one chat out of it, turn off “Use memory in this chat” in that chat's options. Free keeps 20 memories, paid plans 100.

• Projects

  A project gives a set of chats the same instructions and files. Project files can be PDF, Word (.docx), Markdown, text or CSV. Create one with the + beside Projects in the sidebar; a chat you start on its page is in it, and “Move to project” in a chat's options moves one in or out. Free includes 1 project with 5 files; paid plans have unlimited projects with 50 files each.

• How files are used

  A small project's files (up to 60,000 characters of text in all) go with every message. In a larger one, each message gets the parts of the files most relevant to it, up to 20,000 characters.

• Chat length

  Your memory and a project's instructions and files count toward a chat's length, like its messages: past 64k tokens each reply counts as 2, past 128k as 4, and a chat stops at 200k. The composer says so before you send.

• Run code

  The model writes Python or Node.js and runs it in an isolated sandbox after you approve (counts as 1 Claude Haiku 4.5 message a run, up to 60 seconds). Settings → Tools can skip the approval.

• Images

  Ask for an image and the model makes one, with the image model you pick in the Tools menu (from 1 Claude Haiku 4.5 message an image: Qwen-Image 2.1 (Alibaba, via kie.ai): 1; GPT Image 2.5 Flare (OpenAI, via kie.ai): 2; Nano Banana 2 (Google, via kie.ai): 2; Seedream 5.0 Lite (ByteDance, via kie.ai): 2; FLUX.2 Pro (Black Forest Labs, via kie.ai): 2; Ideogram 3.0 (Ideogram, via kie.ai): 2; Nano Banana Pro (Google, via kie.ai): 5).

• MCP servers

  Connect MCP servers in Settings → MCP to give the model more tools. Each tool call asks for your approval unless you allow it always.

• How plans and messages work: plans, messages and limits.

• Where your messages go: Privacy Policy.