Juntang Zhuang (@archanfel

Juntang Zhuang

170 posts

Juntang Zhuang

@archanfel_anoth

prev lead xAI PreTraining, grok 4-fast/4.1/4.2/4.3 and coding rl, grok 2/3/4 core contributor. ex-OpenAI, inventor of GPT4-turbo 128k long context algo

juntang-zhuang.github.io

Joined September 2018

Juntang Zhuang
@archanfel_anoth
Apr 15, 2025
Thrilled that you like it!
Bindu Reddy
@bindureddy
Apr 12, 2025
Grok-3-mini is intelligence that is too cheap to meter! It's just below Sonnet 3.7 thinking, but it is priced highly competitively 1M input - $0.30 and 1M output - $0.50 Cheap and performant models are the future! We need trillions of tokens to automate all work
2.3M
Juntang Zhuang
@archanfel_anoth
Aug 11, 2025
It’s extremely fun though tough to train the first natively multimodal model ever in xAI. Excited to co-lead the pretraining of v7, super honored to be a member of the fantastic team! we are working hard to further improve this model!
Elon Musk
@elonmusk
Aug 10, 2025
Replying to @elonmusk
Our V7 foundation model, which finished pre-training last week, is natively multimodal. It processes a video/audio bitstream directly, understanding it without converting it into anything, so, for example, it will finally understand nuances in how you speak that convey mood and
46K
Juntang Zhuang
@archanfel_anoth
Aug 14, 2024
Grok2 and Grok2-mini is here, xAI speed is insane!
xAI
@xai
Aug 14, 2024
x.ai/blog/grok-2
56K
Juntang Zhuang
@archanfel_anoth
Apr 19, 2025
Replying to @kimmonismus
No, we designed it to be this fast and cheap
14K
Juntang Zhuang
@archanfel_anoth
Apr 22, 2025
It has been really exciting to cook grok3-mini, an amazing experience of collaboration across teams in xAI! Here are some fun facts: 1) The pretraining was a one-shot run, no rewind back, no loss spike, and it's a cost-efficient train. We have a very stable training recipe and
xAI
@xai
Apr 18, 2025
Replying to @xai
Let’s start with Grok 3 Mini. When we set out to build a fast, affordable mini model, we knew it would be good but even we didn’t expect it to be this good. Some highlights: - Grok 3 Mini tops the leaderboards on graduate-level STEM, math, and coding, outcompeting flagship
71K
Juntang Zhuang
@archanfel_anoth
Feb 18, 2025
Excited to be a member of the amazing team at @xai , and shipping the best grok3! Thrilled to lead grok3-mini training, and will ship it to all users for free in the coming days! LFG!
155K
Juntang Zhuang
@archanfel_anoth
Feb 20, 2025
Glad you like it! We are working hard to improve grok3 and will continue to deliver grok4!
Derya Unutmaz, MD
@DeryaTR_
Feb 20, 2025
I must say, I’m starting to like this Grok 3 guy!☺️ One thing I noticed today is that it truly seems to understand (Grok) what it’s talking about in the Richard Feynman way of explaining things: “If you can’t explain something in simple terms, you don’t really understand it.”
23K
Juntang Zhuang
@archanfel_anoth
Aug 16, 2024
🚀 It took me a ton of efforts to do most of Grok2 mini training work, and super excited to see it serving many users, hope you like it! What features should I add next? Drop your thoughts below!
24K
Juntang Zhuang
@archanfel_anoth
Aug 24, 2025
Feel like a decade again ago when we designed and trained your previous generation @grok
Elon Musk
@elonmusk
Aug 23, 2025
The @xAI Grok 2.5 model, which was our best model last year, is now open source. Grok 3 will be made open source in about 6 months. huggingface.co/xai-org/grok-2
37K
Juntang Zhuang
@archanfel_anoth
Mar 25, 2025
We are hiring researchers and engineers at xAI to build next-gen naitive multi-modal models, please apply online or dm me if you are interested!
job-boards.greenhouse.io
job-boards.greenhouse.io
10K
Juntang Zhuang
@archanfel_anoth
Feb 20, 2025
Try grok3 for free!
xAI
@xai
Feb 20, 2025
This is it: The world’s smartest AI, Grok 3, now available for free (until our servers melt). Try Grok 3 now: x.com/i/grok X Premium+ and SuperGrok users will have increased access to Grok 3, in addition to early access to advanced features like Voice Mode
00:00
33K
Juntang Zhuang
@archanfel_anoth
Aug 5, 2025
Replying to @gu_xiangming @SonglinYang4 and @OpenAI
I brought this and other attn variants into GPT two years ago, other ones that make a even larger difference is hidden from this oss model
16K
Juntang Zhuang
@archanfel_anoth
Aug 14, 2025
Replying to @ibab
It’s a great honor to work with you, thanks for all your support and I still remember when we were debugging together in the small bunker room at Tesla. Wish you all the best and a wonderful next journey!
15K
Juntang Zhuang
@archanfel_anoth
Jul 11, 2025
Groot4
6.6K