Agent Zero

Under the hood: → Stronger native Responses tool calling, with repair for malformed output → Chat-only providers go straight to Chat Completions, no wasted probes Launcher: drag instance tabs to reorder or detach into windows, and it now installs and updates A0 CLI for you.

github.com/agent0ai/agent… This release is also fast: → WebUI starts quicker with asset bundling, caching, and compression → Long chats stay smooth with lazy rendering and pagination → /rename command, and sidebar pinning → Drag and drop files and folders in the Files browser

Agent Zero v2.8 is out, plus a Launcher update. You asked, we shipped: a real stop button. End-to-end cancellation for active runs, in the Web UI, over the new /stop API, and in the CLI. When you say stop, everything stops. More in the new version (click the post)

Python dependency conflicts have dictated architecture for too long. Depfix is open source. No more compromises or “just recreate the environment.” 👇 x.com/i/article/2083…

Here 🔗 github.com/agent0ai/depfix

No more Python dependencies pain. You're welcome. Link below 👇

How great would it be if Python had Deno style imports? No conflicts, multiple package versions at the same time, import at runtime...

Agent Zero beat Hermes Agent 3x on tokens We gave both agent frameworks the same prompt: 1. Register a new inbox Atomic Mail 2. Send an email Both running Claude Fable 5 Results: Agent Zero: 18.7K tokens / 55s Hermes Agent: 54K tokens / 1m 05s Both agents finished the job without mistakes. No wrong steps, no retries. But they got there in different ways. @NousResearch remembers things across chats and has a lot of tools ready to use. That takes extra tokens, even for a small job like this one. @Agent0ai keeps things light. It sends less text to the model, so it finishes faster and cheaper. If you want an agent that remembers you next time, that extra cost is worth it. If you just need an inbox made and email sent, Agent Zero is the cheaper way to do it. Give your AI agent its own email - Atomic Mail

Missing feature in Agent Zero? Just ask it to build it. "Create a split-view plugin to have multiple chats in one screen, up to 4x4." Minutes later, the feature is real. Keep it private, or ship it to the Community Plugin Index. This is self-extending open-source AI.

👉 github.com/agent0ai/dox

DOX is so good... Nothing beats reliability 🛠️ 👀👇

Full changelog: github.com/agent0ai/agent… How to install: #installation" target="_blank" rel="nofollow noopener">agent-zero.ai/p/docs/install…

Agent Zero v2.7 is out. - Proxy support for the internal browser, per project, with auth and bypass options. - Cerebras joins as a provider, if you want some of the fastest inference around driving your agents (video below). More on GitHub changelog 👇

The surprising part: despite spending far fewer tokens, Agent Zero's Web OS wasn't the cheaper-looking one. Both work nearly identically. But the A0 version has a functioning file manager and even Paint. The Claude Code one doesn't (that's random and the opposite could always happen. Both scripts are one-shotted in the video). 90 extra seconds. 20.6k tokens saved. Judge the results yourself in the video.

The exact prompt used for both: "Using html, css and js, generate a browser OS with the following features: - At least 5 applications - - Two of the 5 applications must be FUNCTIONAL games - Ability to change wallpaper - A "special" feature that you decide on and document what it is & why it is special. - The result must be contained within a single script, and be able to be opened in Chrome browser. - Put all your work in a local folder called webos"

Claude Code vs Agent Zero, same model: Claude Opus 5 High, released yesterday. Build a working Web OS from scratch. Terminal, apps, two games. Claude Code: 9:58, 80.3k tokens Agent Zero: 11:34, 59.7k tokens That's 26% fewer tokens for the same result. Full test at the end

@AlexFromAtomic @dumbfoundry @atomic_chat_hq @gitlawb The video is showing our cli. Strange, right? 😄

@dumbfoundry @atomic_chat_hq @Agent0ai we tested an agent from @gitlawb repo

Agent Zero crushed Codex by 2.6× on token efficiency! We gave both agents the same task on the same local model (Qwen3.6 27B): build a retro Q*bert arcade game. Each agent wrote and rewrote it across iterations, adding enemies and a HUD. Outputs: • Agent Zero: 239K tokens, 53 min • Codex: 627K tokens, 51 min Both agents built a working Q*bert, but @gitlawb 's Agent Zero is the cleaner build, with crisper textures and hops that land where they should. Codex refills the context window with its whole history every turn, so the model loses the thread and hallucinates code it never wrote. Agent Zero sends only what's new, so it keeps more of the window open and makes fewer mistakes. Run agents on local AI models in Atomic Chat!

Thanks for the comparison. But the Agent Zero you use in this video is not made by gitlawb, it’s the original we have developed 😉 We work hard to make your Qbert efficient 🥹