2x faster then ollama?
What about vs llama.cpp or lm studio?
We just launched our new inference engine! It's over 2x faster than Ollama, without any new hardware
On average across 11 popular local models:
Mac M4 Pro vs Ollama (MLX) 1.32x faster prefill, 2.19x faster decode
DGX Spark vs Ollama (llama.cpp) 1.34x faster prefill, 2.17x faster decode
Ships as a desktop app on Mac, Windows, and Linux
100% free and open source ⬇️
The open source bots are here.. this looks really interesting as well! Competition is good.
Simpler then Hermes? At this point Hermes bots are broken on my setup after each update.
More secure then openbot? Serious security concerns on Linux, Mac see my post.
Who's using Mausbot?
Three releases yesterday. MausBot 0.1.130:
• Simple mode redesigned: calmer, follows your light/dark theme
• New Bot just makes the bot. Say "Set me up"
• Bots answer first, ask one thing, and "Remember" really remembers
• Charts and wide tables right in the message
• Teams: one inbox, your own Chief
GD has entered the .SI chat
Welcome & Liftoff
Say hi to .si
.si is now available at GoDaddy. Super intelligence has a domain extension, so claim your .si domain:
godaddy.com/tlds/si-domain
Antigravity IDE [2.6.0] has Gemini 4 Argon now
Wild, X algo change or just got lucky? My post about openbot went totally viral - 20K+ at this pt.
[teapawt.net/jay_k/status/211470248…]
How?
All I did was post about Hermes vs Openbot [didn't even try it]
@grok any ideas?
I've been running Hermes since day one.. it's been good.. they set the bar for open source agents for sure but there's always been these constant issues..
The UI on the desktop in Linux is clunky have to constantly tweak the config and there's no mobile app that's probably the most important thing
This just looks clean and simple and intuitive
Anyone else using openbot?
Security doesn't look great for @OpenBot_ [not running in isolation/docker] on Linux..
I asked - What does this mean?
"for openbot 'Unrestricted Host Execution: Runs agents with danger-full-access and approvalPolicy: never by default. Prioritizes zero-friction automation over containment."
By default, OpenBot runs AI agents directly on your primary operating system with danger-full-access enabled and an approvalPolicy: never setting. This configuration explicitly prioritizes frictionless automation over system security and isolation.
No Containment (Host Execution): The agent operates directly within your machine's primary environment rather than an isolated container or virtual sandbox. It inherits all rights of your current user session, including access to local networks, system processes, and installed developer tooling.
Unrestricted Privileges (danger-full-access): Safety boundaries are removed. The agent can read, modify, or permanently delete any file your account can touch—such as sensitive source code, configuration files (~/.bashrc), and credentials (~/.ssh keys and API tokens). It also possesses the authority to execute arbitrary shell scripts and install packages.
Bypassed Approvals (approvalPolicy: never): Standard human-in-the-loop safeguards are completely disabled. Whenever the agent determines it needs to run a potentially destructive terminal command, it executes it immediately in the background without prompting you for confirmation.
The Fix: Avoid running OpenBot natively on your main development machine.
Always isolate it inside an unprivileged Docker container, a disposable virtual machine, or a separate remote development environment so it cannot affect your critical local data.
I've been running Hermes since day one.. it's been good.. they set the bar for open source agents for sure but there's always been these constant issues..
The UI on the desktop in Linux is clunky have to constantly tweak the config and there's no mobile app that's probably the most important thing
This just looks clean and simple and intuitive
Anyone else using openbot?










