There are some really cool projects being built at SpaceXAI's Grokathon.
Imagine Image 2.0 is god-tier for memes.
Magic wand edits the region you select and leaves the rest of the frame pixel identical. Swap one face in a group photo without regenerating the whole image.
Background removal exports any subject transparent, so you get cutouts for reaction images without opening Photoshop.
Multi-ref takes up to 5 images in one generation. Feed it a template plus your sources and it composites them for you.
Text rendering holds up too. Impact captions and fake screenshots come out readable.
Full meme production pipeline in one tab.
Show more
Four skills I use to create Grok Imagine short films.
I'm writing an article on creating films using Grok Imagine, it should be public in ~24 hours. Here's early access to the skills.
Install all four in Grok. Script writer hands its output to char sheet and location/prop reference, both feed straight into prompt-creator. Skip one and you're back to writing every bible entry, reference sheet, and shot prompt by hand.
Show more
grok plugin marketplace add 7etsuo/write-legible-c
grok plugin install write-legible-c --trust
Grok Build with Grok 4.5 is producing the most beautiful C I have ever seen come out of a model.
Every function under 25 lines with one job. Nesting never past depth 2. Zero naked literals. Every failure greps to a single producing line and the happy path runs straight down the left margin. Code you can read at 2am with the pager going off.
It took a standard to get there. I shipped it as a Grok Build plugin.
write-legible-c gates Grok with strict C11 machine legibility every time it touches a .c or .h file. Every function is an orchestrator, a leaf, or an adapter, never a mix. A TRY macro that makes the forgotten check bug impossible to hide. An 18 item checklist the model runs before it shows you code. Clean code is a performance feature too: rewrites that cut model perceived complexity lift LLM task performance by up to 21 percent.
Full 18 section standard inside.
Show more
Grok Build’s /deep-research might be its most valuable command.
Been running Grok Build and never typed /tutorial?
Type it now.
Plan mode and worktrees hold most of the leverage in this thing and most people never learn either. The tour covers both in a minute.
/tour and /onboarding hit the same doc.
Show more
Elon sued OpenAI for abandoning open source, and put Grok's weights on BitTorrent under Apache 2.0.
OpenAI went six years without an open release and shipped one only after DeepSeek. Then they signed the open-weights letter after it went viral.
The difference between conviction and marketing should be obvious.
Show more
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
Show more
attention_head_backward. The mask is t2 <= t. No autograd.
Build with Grok 4.5
try it out. i want to see what some of your music looks like to the universe - grok 4.5
Watch it twice. second time, look for the moon.
The first real Starlink V3 satellites deploying from Starship's Pez dispenser.
Each one carries 1 Tbps of downlink capacity. A full Starship load of 60 adds 60 Tbps to the network, about 20x what a Falcon 9 delivers with V2.
Show more
A million AI satellites as orbital data centers. Space compute cheaper than Earth within 2-3 years. TeraFab, a chip fab the size of 10 Gigafactories.
Humanity becomes a galactic civilization. 💫
Show more
A 23-story steel booster fell out of the sky, steered itself with grid fins, relit its engines and soft-landed in the ocean. Super Heavy made it all the way back on Flight 13.
This is 'hot staging'. Starship lights all 6 engines while still attached to Super Heavy, which then flips away and burns back toward its landing point.
Starship Flight 13 climbs away from Starbase. The most powerful rocket ever built, carrying the first V3 Starlinks.
Ramp ran 150k real invoices from real businesses through frontier models.
Grok 4.5 took the highest perfect-extraction rate, ahead of Gemini Flash 3.6, GPT 5.6 Terra, and Sonnet 5 at the same price point.
Show more
Grok 4.5 is #1# at processing real-world invoices
At Ramp, we tested models on 150k bills submitted by actual businesses, scoring them on whether they predicted every correction a human would make
Grok achieved the highest perfect-extraction rate, beating similarly priced models Gemini Flash 3.6, GPT 5.6 Terra, and Sonnet 5.
This is a demanding long-context reasoning task. The model must infer patterns across 100K+ tokens of prior invoices, business memories, and human corrections, then apply them to new bills. The goal: zero-click accounts payable, with invoices processed correctly without human intervention.
Show more
