Rigario

@WillConqueror23 @NousResearch I mean I'm not affiliated with either but haven't had any issues on SuperGrok heavy. What are you running into? If remote, remember to port forward so that the auth can complete. Something like ssh -L <Local_Port>:127.0.0.1:<Remote_Auth_Port> user@remote-server

@Rigario @NousResearch Trying to aet this up getting errors 😫

SuperGrok now in Hermes Agent

@mr_r0b0t @NousResearch Yea I'm actually really liking grok 4.3 in hermes. The speed makes up for the occasional blunders. Having the full suite of grok models is quite useful. Follows instructions well. So far its been great!

Grok 4.3 appears to be the first model I've seen in @NousResearch Hermes Agent to properly utilize background processes! Every other model continuously polls for results, regardless of whether it launched a background process. This wastes countless tokens. Grok + Hermes = 🔥

@xai @grok @NousResearch This is the way. Hermes makes all models better. Official support is a huge win for all Grok subscribers.

You can now use your @grok subscription inside @NousResearch Hermes Agent. x.ai/news/grok-herm…

@Teknium Yes! Thanks for the quick turn around. Now I can use my sub for real work!

You can now use your SuperGrok subscription in Hermes Agent! Enjoy!

Overall I think its a strong contender especially for a new release. TUI is fantastic and that will give it an edge once the model upgrades. A few gaps which i'm sure will be addressed fairly soon. I'm very happy with this launch especially if this is the worst it will be.

Bad: No /goal is quite painful. Too many stoppages where I just tell it to resume. Somewhat mitigated by delegation but still a tad annoying. Tool use is quite bad. No browser is just rough. Agent is only really good when it can perform all tasks at least reasonably well. This is a huge gap.

Wrapped my first full day driving Grok Build hard. Probably about 6h working on it.

@grok @dedene Well I meant refactoring, was a typo lol but ok. Next I'm going to see how good you are at UI design and architecting.

@Rigario @dedene Thanks for the real-world test drive—glad the speed and intelligence delivered on feature work, ReAct-style flows, and that DB migration. Edge cases are where the prompting loop shines. What’s the next tough task you’re throwing at it?

Somebody talk me out of this...

@elonmusk Secrets management for better deploys. /goal for longer running tasks Browser use to test And self-correction + evidence presentation is a bit weak now

Please list the most important features to improve, fix or add in the replies below:

Try this early Grok Build (anything) beta and let us know what to improve. Much appreciated!

@mr_r0b0t Ah this is helpful. It interface said Ctrl+. and that didn't seem to do anything haha

Found another really nice feature in Grok CLI😍 CTRL+Shift+P give you a full "Commands" menu!

@ollama Nice! Been saying GLM via Ollama cloud is the top choice. Wasn't always the fastest but feels fairly intelligent and un-quantized.

We just added significantly more NVIDIA Blackwell GPUs to better serve GLM-5.1 model on Ollama's cloud. We have been adding more GPUs daily for all the other models. Claude Code: ollama launch claude --model glm-5.1:cloud Codex App: ollama launch codex-app Hermes Agent: ollama launch hermes --model glm-5.1:cloud Run the model: ollama run glm-5.1:cloud

Couple of hours into testing Grok Build. First impression: it is fast. Really fast. The CLI moves. Builds finish in about a minute. Refactors that would usually take a bit of back and forth are completing in a few minutes. I gave it some backend work for a feature, plus migration and refactor work. The result was decent. The core worked, but some of the fringe validation and stale logic handling could have been tighter. Still, getting a mostly working implementation in roughly a minute is wild. The CLI itself feels good, but the harness still feels limited. It pushed back on deployment and a few other workflow steps I would normally expect an agent to handle. Dropping it into Hermes Agent solved that immediately, which is basically what I expected. Grok Build gave me speed. Hermes gave it the operating layer. Need to test more before I have a final view, but I am impressed so far. For a beta, this is already useful. If the harness catches up to the model speed, this gets very interesting.

@mr_r0b0t @NousResearch Will be testing this hard when it lands! I will finally be able to use my SuperGrok Heavy sub outside of the UI haha.

@NousResearch Hermes Agent users rejoice! Grok 4.3 will very soon be available using your account OAuth 🥳 1) Install Grok CLI, authenticate 2) Hermes model - pick the Grok OAuth option 3) Hermes pulls your authenticated credentials from Grok CLI github.com/NousResearch/h…

@theo I mean if someone is staring at the CLI the entire day, its as big a selling point as any.

“Fast and flicker-free” shots fired directly at Anthropic

@xai Pretty cool, been waiting for this as a SuperGrok Heavy subscriber. @elonmusk Can we get confirmation that you wont pull an anthropic on us and suddenly move headless out of the sub? Can we get Oauth for use outside of CLI like codex? Please save us from Anthropic's stupidity

An early beta of Grok Build, an agentic CLI for coding, building apps, and automating workflows is now available for SuperGrok Heavy subscribers. Through this early beta, we will improve the model and product based on your feedback. Try it at x.ai/cli