ZeroLog
Private preview
EN DE

264 entries · August 2026 – November 2026

Newsroom

Product changes, pricing and privacy updates, and what each one changes for you. Newest first, with release notes and links to press and directory listings.

Subscribe to the feed Press and listings

Filter by category

Gray to violet: background to important changes. Editorial user impact, not security severity.

  1. Product

    A lighter chat start screen

    The animated chat background uses less graphics work, especially on notebooks with integrated graphics. For people using the animated chat start screen, especially on integrated graphics: less background graphics work and automatic slowdown on struggling devices. No action needed.
  2. Product

    The crystal logo is now the browser icon everywhere

    Tabs, bookmarks and the home-screen icon on your phone now show the crystal ZL logo. Before, the site and the app showed two different, older icons. The tab, bookmark and home-screen icon of the site and the app is now the crystal ZL logo. Features, texts and prices are unchanged. No action needed; a hard reload shows it sooner.
  3. Product

    The first chat message appears while the chat is prepared

    New chats show your pending message sooner, and switching conversations keeps the chat page open. Chat users see their first pending message sooner and switch conversations on the same page. No action needed.
  4. Product

    3 November: a press page that says what each mention is

    ZeroLog now has a press page that separates one interview from four directory listings we filed ourselves, with the same labels in the structured data and in llms.txt. Adds a press page that labels one interview and four self-filed directory listings for what they are; nothing changes for people who use ZeroLog.
  5. Product

    3 November: a documented public API, with the spec checked on every deploy

    ZeroLog now has a public API reference at /docs/api and a machine-readable specification at /openapi.json. Both cover the endpoints you can call from your own code, and nothing else. Adds a public API reference at /docs/api and /openapi.json for the model, token and webhook endpoints; nothing changes for people who do not script against ZeroLog.
  6. Product

    3 November: the app takes on the look of the start page

    The signed-in app now uses the same near-black ground, Barlow type and hairlines as zerolog.ch. Colour shows up only as light: on the open chat, the message field and the send button. In the dark theme the signed-in app looks calmer and matches the start page. Features, texts and prices are unchanged. No action needed.
  7. Product

    Replies now move to the new server during an update

    A routine update cut off a long agent reply on 2 November. Running replies are now paused at a safe point and handed to the new server, which finishes them while your chat follows along. Anyone chatting while an update rolls out. A running reply, including long agent runs with tools, now moves to the new server and finishes there; the open chat follows on its own. No action needed.
  8. Product

    3 November: ZeroLog Setup 0.2.1 without admin rights, web search works again, sharper chat start view

    The Windows setup now installs for standard users, the chat can use web search again, the chat start view draws its fibres sharp, and the Firefox listing is linked.
  9. Product

    Your AI. Your Tools. Your Freedom.

    ZeroLog has a new brand message: your AI, your tools, your freedom. One workspace, thousands of models, zero vendor lock-in. The start page headline, title and social cards carry new wording. Features, prices and plans are unchanged. No action needed.
  10. Product

    2 November: advisor for every model, idle containers pause, waitlist invitations, and housekeeping

    Ultracode agents can ask a stronger advisor model, idle containers are paused again, waitlist invitations work from the admin area, and several internal changes shipped without a visible effect. Ultracode users get an advisor tool, idle containers pause, and invited waitlist people get in. Everything else is internal. No action needed.
  11. Security

    2 November: a hard disk limit per workspace, and the editor on its own address

    Every workspace now has its own disk limit with usage shown in the container view, and the web editor of a container runs on a separate address that carries none of your site session. If you use containers, you now see disk usage and a limit and get a warning at 90 percent, and the editor opens on a separate address. Check your usage; no other action needed.
  12. Product

    2 November: Research searches the web for real, grades its own report, and always ends with a document

    Research now runs real searches, accepts only pages with a finding as sources, has a reviewer grade the draft, shows a progress card, and opens the report in the side column. If you use Research, reports rest on better sources, show a grade, can be downloaded in five formats, and no longer end without a result. No action needed.
  13. Product

    2 November: the agent can work inside Windows VMs, and new VMs answer SSH sooner

    The chat agent can now run PowerShell commands and read or write files inside your Windows VM over SSH. New VMs reach SSH in minutes, and a prepared machine makes it faster still. If you use Windows VMs, the agent can now work in them, and new VMs become usable sooner. Without VMs nothing changes. No action needed.
  14. Product

    2 November: instant chat switching, a model picker that opens at once, tools on by default

    Switching between chats no longer waits for the server, the model picker opens with its list, new chats start with tools on, and chats that had broken on a missing tool result work again. Chat feels faster, new chats use tools by default, and chats that failed after a missing tool result recover on the next message. No action needed.
  15. Security

    What five Strix security scans found, and what we did about it

    Five scans on 28 October produced 63 distinct findings, 2 of them critical. Most are fixed in code; one is contained rather than solved, and a re-scan is next. Security hardening behind the scenes. No action needed; no misuse from outside was found.
  16. Product

    1 November: the Crystal look reaches every page

    After the chat, the rest of the site and the app moved to the Crystal design: docs, legal pages, newsroom, access page, waitlist, jobs, files, billing and the admin overview. Almost every page of the site and the app looks different. Features, prices and plans are unchanged. No action needed.
  17. Product

    A crystal workspace for chat

    The dark chat workspace now uses the Crystal design, with a clearer composer and matching tool panels. The dark chat workspace, composer, sidebar and tool panels look different. Your chats and settings are unchanged.
  18. Product

    30 October: sturdier chat, cheaper Claude calls, own keys in ZeroLog Code

    Six chat faults were fixed, Claude requests through ZeroLog Code can now read from the provider cache, and an own API key in ZeroLog Code no longer draws on the pool budget. Chat, attachments and ZeroLog Code behave more reliably, and an own key in ZeroLog Code no longer runs into the pool budget. No action needed.
  19. Product

    Container start shows a real percentage, Running only when ready

    The Starting view now counts real steps, and Running appears only after a readiness check. On a cold VM that is 41 seconds with no early Running. Starting a container shows real progress, and Running appears only once it is ready to use.
  20. Product

    Sub-agents for cheaper chats, and "ultracode" as a one-turn word

    The chat agent can now hand mechanical work to cheap sub-agents without Ultracode. The word "ultracode" allows workflows for one turn instead of switching them on for good. Chats can get cheaper through sub-agents, and the word ultracode now applies to one turn only.
  21. Product

    Tailscale one click: cancel and try again no longer breaks

    Cancelling and clicking again while a slow container starts now works. The view shows what it waits for, and the tab always says why. Cancelling and clicking Tailscale again while a slow container starts now works instead of closing the tab.
  22. Product

    In the media: Let's Data Science on why hiding agent API keys is not enough

    Let's Data Science published an interview with our founder about key placeholders, approvals and goal checks. The outlet did not test the service. No change to the service. This is a press mention.
  23. Product More login addresses open the sign-in window, the terminal waits briefly for a freshly started container, and a Pro account sees a plain plan hint for own machines. More login addresses open sign-in, and the terminal waits briefly after a container starts.
  24. Security

    Passkey confirmation for vault changes, and use only for databases

    Changing where a secret may go, or showing it, needs your passkey, checked by the key service itself. The key service also reports every change to your protection to you directly. Vault changes and showing a secret now need your passkey, and databases work with use only.
  25. Security

    Credentials in tool output are blacked out before storage and before the model

    A command like env could print a proxy password into the chat. Output of tools now passes a blackout step before it is stored or sent to a model provider. Passwords and keys printed by commands like env are masked before storage and before they reach a model provider.
  26. Product

    A pasted file is no longer left behind when you press Enter

    If you pasted a long text and pressed Enter at once, the message could go out without the file. The chat now reads the attachments at the moment of sending. Pasting a long text and pressing Enter now always sends the pasted file with the message.
  27. Product

    Ultracode agents follow your permissions, and you can see what they do

    Agents of an Ultracode run now run only what the approval level of the chat allows, and a fixed line above the message box shows their commands, errors and requests while they work. Ultracode agents now ask before risky commands and a status line above the message box shows what they do.
  28. Security

    Vault entries now have three states

    Visible, use only or locked: the AI reads a value only where you release it, or works with a placeholder and never sees the real one. Each vault entry is visible, use only or locked, so the AI can work with a placeholder and never see the real value.
  29. Product

    Helper jobs now run on your own key

    With your own provider key, summaries, /goal checks, Deep Research helpers and similar jobs now use a small model of that provider instead of a platform model. With your own provider key, helper jobs now run on a small model of that provider instead of a platform model.
  30. Product

    The Stop button now stops what is running

    Stop used to end the chat view at once while a tool, a workflow or a /goal kept going. Now it reaches all of them, and the button says so until the server confirms. The Stop button now ends running tools, goals and workflows, and shows Stopping until the server confirms.