Luke Haines retweeted
2022 was the year of autocomplete
2023 was the year of chat
2024 was the year of the AI IDE
2025 was the year of coding agents
2026 was the year of agent orchestration
2027 will be the year of outcome engineering
Gemini 4 Neon has entered the arena
teapawt.org/demishassabis/status/2…
Gemini 4 Neon even surpassed Fable 5.2 in coding agent index and sits 6 points ahead of the yesterday released GPT-6.2 Astra.
30
Luke Haines retweeted
Replying to @GergelyOrosz
Pretty well deserved by Theo tbf, early to market, focussing on execution, and has a good immediate feedback loop with his customers via Twitter
Truly exceptional work. Renting GPU’s blew my mind the first time I read it but it makes so much sense
The Fable of Clawd
A 2 minute 4K film made entirely in Blender, from one prompt on Fable 5.2. It took about 4.5 hours.
Claude built every asset in Blender by writing Python:
- 5 characters and props: Clawd, his friend Pip, a toy truck, a 3D-printed tulip and a gift box
- 7 full sets: a study, a staircase, a street, a workshop, an airport, a night sky and a house
- every camera move and every frame of animation across 52 shots
That came to 12,500 lines of code. It started 7 copies of itself to build the sets at the same time.
Then it rented 89 GPUs through RunPod to render 3,220 frames as fast as possible, 36 of them running at once. Total cost: a little over $20.
My biggest takeaway is my Claude Code knows it has access to cloud compute, so it's much more ambitious.
I tried a similar prompt on my other computer that doesn't have the runpod skill setup and it wasn't nearly as high quality
Full details linked below.
12
Luke Haines retweeted
Fable 5.2 vs. GPT-6.2 Astra - ROUND 2
- 242 KB - 11.1 MB
- $158.88 - $39.68
now with eric's 3D skill + xhigh, it computed baked binaries from .cpp files
once my tokens reset I'll run another benchmark with modified prompt, have some ideas... 🤔
Luke Haines retweeted
I almost forgot my favourite GPT-6.2-Astra generation - this is SVG (yes SVG!!) of a peacock - no other model came close to the precision, lifelike nature and detail of this generation, incl. Fable
Luke Haines retweeted
Stay AWAY from GPT-6.2 Astra on low
If you're dropping Astra to low to save usage, just use Sol and turn reasoning up
I recommend Claude over OpenAI models right now anyway
But if you're going to use OpenAI models, don't use Astra on Low
Luke Haines retweeted
every. single. marketing. department. will. be. flipped. upside. down.
Luke Haines retweeted
the @every team's full ratings for Astra and Sol 6.2
hard to remember a frontier model release like this. opus 5 is the closest comp
every.to/vibe-check/vibe-che…
Luke Haines retweeted
Replying to @nicksaraev
Google has an Ai branding issue.
Also, it's voice models are still not SOTA so nobody cares, nor has the google Ai subscriptions - so they don't get used
Luke Haines retweeted
ChatGPT and Google give you 20x the usage for 10x the money. Claude gives you 10x.
Luke Haines retweeted
If you think Jensen resigned because of the affairs you’ve failed the IQ test of the week
Luke Haines retweeted
This slide from the Anthropic IPO roadshow was a big factor in my decision to leave. It’s an insane plot and I’m surprised they made it public.
It shows the compute efficiency of pretraining over time. The y-axis is a **log scale**: they are seeing continued superexponential algorithmic progress.
Mythos was not a one-time deviation, but a point on a superexponential. They could already retrain Mythos with a fraction of the compute originally needed.
It doesn’t strictly imply a software-only intelligence explosion- they have been spending more on R+D. But algorithmic improvements do not require ridiculous creative genius and are on the verge of being automated. You do the math.
Luke Haines retweeted
why am I just now hearing out about this seedance hack to get around the likeness policy error...
apparently we just need to have Fable 5.2 add a scratch pattern to faces? has anyone else tried this
“A world model for physical intelligence that predicts the next world”
The future is going to be crazy.
Imagine in 5 years that you have models predicting every real life outcome… and then imagine we’re not in a simulation already
This is a MUCH bigger deal than people realize.
MBZUAI's Institute of Foundation Models just unveiled PAN 2, a world model for physical intelligence that predicts the next world: give it a scene and an action, and it simulates what happens next.
So you can ask it "what if?" Take a Jenga tower and pull a block from the middle, and it falls. What if you slowly pull one near the top instead? Run it again, and the tower stands.
But here's where the team really blew my mind: "what if?" again and again.
And you get an environment where an AI agent could learn: try an action, see the simulated result, adjust, try again.
In IFM's tests, an agent even discovered a property of the world it couldn't see, purely through simulated interaction. Try first in the simulated world, act in the real one after.
This is where world models should be heading: not just generating what the world looks like, but learning to predict how it behaves.
And it's open under Apache 2.0: weights, training code and inference code.
This is absolutely nuts.
75



















































