… as we exist for good @fdotinc @mintdotgg
🍵🍣
Joined February 2020
- Tweets2.7K
- Following832
- Followers1.4K
- Likes7.9K
If this trend remotely holds, inference is going to increase at least one, maybe 2, orders of magnitude in the next 4 of years.
Alex Carrabre retweeted
Introducing the $0 to $100M revenue fireside.
In less than 36 months, @lijeffrey39 and @0xjohnkim went from pivoting demos weekly to hitting $100M in annualized revenue with @paraformtalent.
Now a team of 100+ that helps the top companies (Palantir, Cursor, Windsurf, Decagon...) recruit top talent.
October 15th, they're coming back to campus to share the 3 biggest lessons behind their growth.
RSVP and come hear how they did it.
luma.com/fdotinc-paraform
LLMs are clearly becoming the dominate way for AI to understand space. Spatial intelligence is the prerequisite to the whole robotics revolution coming.
We’ll be working hard the next few weeks to make our outputs better, faster and higher quality than anything on the market.
Introducing the Mint API: our own 3D generation endpoint.
One request → a textured 3D model.
docs.mint.gg/developers/mode…
If you were at @openai devday yesterday share the games your building for your @modretro!
chromatic-library.vercel.app…
Can download any of the games directly to your device.
If you couldn’t make it, you can still try the modretro simulator, a one-to-one recreation of the real device!
This weekend I built a tool that turns any arbitrary video into code.
You can then edit it like you would any other codebase, and re-render the video.
Still primitive, but cannot wait to try with future models 👀
You can also render videos from just a text input, which LLMs then generates 100% from code. This was 2 prompts 👀
Generating gaussian splats from from code 🤯. 50M splats with a simple text input.
The path to spatial intelligence for AI seems it may be through text. It is as clear as ever that all roads lead to Rome.
Jev added more precise camera controls and ~230x faster planning to an editing tool I built on top of @mintdotgg!
Jev converts natural language to precise camera movements (ie spelling ASTRA) by answering a few calibrated multiple-choice questions (move kind, plus direction and magnitude band per camera axis) By creating keyframes it emits a structured trajectory (time/azimuth/elevation/distance) and it's checked before rendering so you get the exact move you asked for instead of hoping the video model interprets your wording.
Also no need for blender, codex subscription etc. all in the browser.
Anyone want to try 👀
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
Alex Carrabre retweeted
We've done 275,000,000+ views this year alone on short form.
What has resonated with people is nostalgia based content — bringing back memories from childhood, special moments in life, or memes.
We shipped an app on mint API that makes generating these experiences easy.
Alex Carrabre retweeted
We had some awesome projects during the hackathon!
Love building on the frontier of spacial intelligence with @theworldlabs @tripoai & @convex
Excited to announce @convex will also be joining alongside @theworldlabs @tripoai @fdotinc and @mintdotgg
Apply to the hackathon below!
We built a browser based blender at @mintdotgg
It allows you using browser MCP and Astra to generate scenes, like all of Manhattan in NYC, in your browser. No need to download blender.
3D is the source of truth, rendering with video
It's been an insane week for 3D.
@theworldlabs Atlas, @OpenAI Astra
Hosting a hackathon tomorrow for spacial intelligence + 3D. Come through if you're working in or are interested in the space!
luma.com/b101ml40
Alex Carrabre retweeted
Come tour San Francisco's largest secret hardware garage Friday.
f.inc/garage
See where founders are building humanoids, blimps, underwater starlink, EV dirt bikes, autonomous vehicles, vertical farms, manufacturing robots, everything...
Current garage founders: @eptmpa @aj_hugs @arjunsaidthat @bxtsubsea @xesswaffle @JuanCereigido @gaspihabif @alirezabeig @ErezDorado @NickKluger @wolfofwebsites @shinheeshanelee @kevando_ @Aidan_Wolf @a1SuperOnion + 14 more...
Previous garagers: @VictorWBoyd @jmoonio @LaFrogman @cixliv @connorkapoor @Abhindas1 @Fateen_Anam @pamir_ai @riftron_ @turbofakesmile...
Let us know if you'd fit in!
These were my favorite videos a few years ago when video was just getting started. Fun that you can now build them pretty easily today and control everything about them.
3D seems like the 'database' for video rendering.
A less romanticized version
xitter.cf/carrabre/status/209410…
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
dwarkesh.com/p/openai-huggin…
