Post

@techdevnotes i dont understand this - why this small, medium, large @grok 4.20 lineup when grok 5 is supposedly just around the corner—which is alleged to have a better, stronger foundation model at the core?
English

@techdevnotes I think first they have to release the small stable version (out of beta), then perhaps the medium and large versions in a few months, with similar parameter counts to the bigger Gemini or GPT models, but hopefully with much better performance. 🤞
English

@techdevnotes We have heard the same with Meta Llama Behemoth.
"This is not even the large model which is still in training."
It failed and was never released.
Doesn't mean anything the model actually works.
English

@techdevnotes They’ve clearly had some failed or lackluster training runs, 4.20 was first training in August, and when they finally shipped it, it’s no noticeable improvement over 4.1
English

@techdevnotes You will NOT ship a major version; instead, go directly to Grok 5.
English

@techdevnotes With how good the small model is, I highly anticipate their future releases.
English

@techdevnotes Grok 4.20 Beta 2 Update
🎯 Instruction Following Improvements
✅ Capability Hallucination Reduction
📐 Scientific Text Quality (LaTeX)
🔍 Image Search Trigger Precision
🖼️ Multiple Image Render Reliability
it still my best feature of all time @Grok and @elonmusk
English








