Jacob Gold
@jacobgold
origin @ graphite @ cursor @ xai @ spacexai @ spacex (soon)
428 Following 1.8K Followers
my body is a machine that turns bad x content into worse x content
one day maybe i'll accomplish this feat
BREAKING: A non-mathematician lab employee used an UNRELEASED model to narrow the P vs. NP problem down to just TWO possible answers
guys
proof of 0 = 1 in Lean, exploiting a hash collision (weak 32-bit hash Expr.hash/mixhash)
incredible
Grok 4.6 was the first model trained on internal model-development tasks.
My team built the training and evaluation stack that enabled Grok to learn from work that accelerates model development itself, including production inference and kernel optimization.
Grok 4.6 now leads our internal MTS Eval and InferenceEval. In one experiment, we gave an earlier checkpoint our production inference codebase in a fully automated environment. It explored 297 optimization ideas and shipped 3 changes on top of an already human-optimized stack, improving prefill throughput by 3.1% and decode throughput by 1.5%.
Our work is featured in the R&D Enablement section of the Grok 4.6 model card.
Show more
50% off a brand new set of patagonia outer layers to replace my 15 year old ski gear? absolutely
my grok bot will see you there
@sshell_ there's noise for sure, but in my experience it's tracked pretty well to real world performance!
BTW you used to compete on the rs-sdk demo server, is that right?
We have a king of the hill thing going on in the wilderness right now if you want to spin up a swarm
Show more
alright you can bot osrs but not on jagex servers
Grok 4.6 is Runescape bench's new #1#, at only ~60% the cost of previous leader Fable 5.
we are moving very fast, with a very lean team. coming up to grok bot mobile next week:
- ipad coming! @7cezmusic is cooking.
- a superb chat scrolling experience better than native.
- directly sharing image from other apps to your bots.
- a load of fixes.
all built with @bot
Show more
the grokbot little guys are so cute
We've released the model card for Grok 4.6!
It goes in-depth on the capabilities of the model across many different evals for coding, engineering, knowledge work, and more. We also discuss pre-deployment safety testing and our safeguard stack.
Show more
small but mighty
The 1.5T that could!
Bigger and even better models on the horizon.
I’ve been using it for ~2 hours and I’m absolutely blown away at how I could automate a bunch of my workflows. Cursor team built a way for “normies” to automate stuff. Feels like an OpenClaw 2.0 moment.
(Zero affiliation, I upgraded to Ultra to use it and absolutely worth it.)
Show more
Grok 4.6 just dropped, and the Fast variety is pretty impressive. Runs really fast at $4 in/$12 out! Compared to other Fast modes, it's about a quarter of the price. Started it out with just a simple PR. No notes!
Show more
does it live up to these numbers? let us know
Introducing Grok 4.6.
It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price.
