Docs
Changelog
New features, improvements and fixes, newest first.
October 17, 2026
New
- File attachments in chat. Attach text and code files (up to 512 KB) or images (up to 5 MB, for vision models) with the paperclip, by dropping them on the message box, or by pasting. Guide
- Export chats as Markdown, PDF, CSV or JSON from the chat header. Guide
- Switch models mid-chat. Click the model name under the chat title; earlier replies keep the model that wrote them.
- Rename chats from the header, or by double-clicking a chat in the sidebar.
- Save a chat message as a prompt, with the model’s reply saved as its first response. Guide
- Usage chart. A line per model for tokens, cost or requests per day, with 7-day, 30-day and all-time ranges. Click a model to hide or show it. Guide
- One Models table. Every provider’s models and prices in one place. Add a model to a provider and it appears in chat right away. Guide
- Add response dialog. The + button next to a prompt’s responses opens a dialog to paste a response and pick its model and prompt version.
- Favicon for the app.
Improved
- You can scroll up while a reply is being written; Follow reply jumps back to the end.
- Providers that reject browser requests are marked Needs CORS proxy, and their errors say so instead of a generic message.
- The New chat dialog lists the provider’s models in a dropdown, with the first one preselected.
- Responses saved from chat are priced with the chat provider’s price for that model.
- 7-day and 30-day usage ranges show every day, including days without usage.
- The Save as prompt form explains the placeholder syntax.
- The sponsored slot now also appears in the chat sidebar.
Fixed
- Usage chart labels no longer overlap or fall outside the chart.
- The chat header’s model picker is now as wide as the model name.
October 16, 2026
First release
- Prompt library with 1,000 built-in prompts and example responses, search, and filters by type, category, keyword, favorites and responses.
- Typed placeholders (text, number, file and select) and Generate, which turns them into a form. Reference
- Versions and responses for every prompt, with input and output cost per response.
- Token pricing: estimated cost per prompt and response, priced with the model of your choice.
- Add Prompt and bulk JSON import.
- Chat with 28 free LLM APIs using your own keys, with Markdown replies and tokens, cost and time for every message.
- Local-first: prompts, chats, keys and prices stay in your browser.