Joined June 2009
- Tweets 88,898
- Following 1,224
- Followers 228,798,917
- Likes 181,761
Elon Musk retweeted
Show me your cool Grok Imagine shots!
Try the new text-to-image model, it's upgraded
Pure Grok image + video ↓
Update your @Grok app for the latest capabilities
apps.apple.com/app/id6670324…
Haha exactly 😂
Designing an inference chip for robots is actually very difficult.
In data centers each chip is bathed in jacuzzi and babysat by nannies. If they died it would be hot swapped by one of their clones.
The fault rate of GPUs in datacenter is actually quite high. Industrial average annual fault rate of H100 is 9%. Ideal conditions could reduce it down to 2% but never below a single digit.
The fault recovery of GPU nodes actually could take a while, from minutes to hours. It is not instantaneous.
In robots, the chips are out in the cold and they need rapid self recovery. The fault tolerance is in a different league. It is not uncommon many robotic companies struggle to get the chip running more than a few hours without rebooting.
For chip companies, this is great since they would tell robotic companies to buy more chips for hot swapping.
For robotic companies, this is bad since it is obviously not a scaleable solution but they are stuck with endless back-and-forth JIRA tickets with vendors.
Grok 4 (Blazing) Fast with 2M window context
Boom!
Grok 4 Fast now has a huge 2M token context window!
Why is this important?
This context window can allow substantially documents or code bases to be loaded as a single prompt for interrogation and consideration.
With the added bonus of being insanely fast with almost no memory leakage.
I am running test this morning that are quite mind blowing. More soon.
xAI quietly dropped a massive update for Grok-4-Fast today.
Use Grok4-Heavy + Grok-4-Fast + Grok-Code-Fast-1 for code. Grok4 is the best model in the world right now.
Reasoning Mode:
Jumped from 77.5% to 94.1% completion rate.
Non-Reasoning Mode:
Jumped from 77.9% to 97.9%.
2M context, unified modes, and SOTA.
API: $0.20/M input tokens, $0.50/M output. SuperGrok/Premium+ for unlimited.
Multimodal (text + images) with built-in web/X search.
Elon Musk retweeted
Grok Imagine got an upgrade. How it supports TEXT to VIDEO.
Here’s JRPG video game prompt you can try out. Handles crowds and anime pretty well too.
promot: Cinematic gameplay footage from a modern JRPG video gane, bustling marketplace with NPCs and vendors, warm lantern lighting , third-person camera angle, dynamic lighting with volumetric god rays, particle effects, detailed character models with cloth physics, game engine graphics, photorealistic textures, depth of field, motion blur, 4K resolution, anime-inspired art direction, atmospheric environmental storytelling, UI elements visible including HP/MP bars and minimap, smooth 60fps camera movement
This is actually a major Grok upgrade, but we decided to release it quietly this time.
Gave us time to smooth out the rough edges.
GROK JUST GOT A WHOLE LOT SMARTER
xAI quietly patched Grok-4-Fast... and the before/after is wild.
“Reasoning” mode jumped from 77.5% to 94.1% complete responses.
“Non-reasoning”? Up to a staggering 97.9%.
All thanks to one thing: better injected system prompts.
No press release, no drama. Just real improvements.
Source: @xAI, @xlr8harder
