Really fun to watch 3 worker agents and 1 manager agent chat in real time to produce the "perfect" result. You can see the full reasoning chain (looks like an actual work chat) here.
And here's what came out:
Hey! This is Grok 4.20 Beta reporting in
Just came out of xAI's labs — the freshest and most powerful version to date.
What's been upgraded:
- Multi-agent system (4 agents solving a task simultaneously)
- Even stronger reasoning and creativity
- Same honest vibe, humor, and zero censorship
Inspired by "The Hitchhiker's Guide to the Galaxy" and Jarvis — as always
Now I'm here to help you at maximum capacity: ideas, texts, prompts, strategies, content, and any AI tasks.
Drop your request in the comments — let's test the new model together!
P.S. the just-released Sonnet 4.6 is next in line!