Product
A new start page, a quicker app and large attachments that fit
zerolog.ch has a new start page. Settings toggles respond at once, and big text attachments stay in your container instead of filling the chat context.
The start page is new and the app responds faster, with nothing for you to do.
The start page at zerolog.ch is new. The four products circle the ZeroLog mark, and the intro animation plays only on your first visit. While the app loads, you now see an outline of the chat instead of a "Loading" text.
Many small things got quicker. Switches in settings (AI tools, image generation, memory pins, the grounded-answers toggle) change on screen immediately and roll back if the server refuses. Preferences are saved after 150 ms instead of 600 ms. The model list loads from a cache and refreshes in the background. Reloading a chat no longer flashes an empty new chat first.
Text attachments above 100,000 characters that already sit in your container are no longer pasted into the prompt. The model gets the path, the size and an outline with line numbers, and reads the file in parts. read_file takes an offset and a limit, and a new search_file searches one file. A 2 MB plan used to be about 1.2 million tokens, more than a 1M-token context can hold.