for the past 4 days me and Fable 5.2 and I have been building an underwater stranded survival game.
everything you're seeing is rendered in engine. the ocean surface, the raft, the reef, every fish. it's all real time, not pre-rendered, all of the code done with AI
you wake up on the torn-off half of a submarine. your stuff is scattered across the sea floor, there's a wreck down there, things that stare back at you in the dark.
There's actually quite a bit of lore I wanted to layer out through the game, all of which you can't see here, but I think I was able to craft a pretty interesting story behind what will be a playable experience.
Alibaba Cloud's Qwen4-Max vs DeepSeek V4 Pro: a pelican riding to the bakery through Lisbon, built entirely in code.
Qwen4-Max delivered a full little world: a feathered pelican, tiled streets, shops, a tram and a control panel for the camera and the light.
DeepSeek V4 Pro's pelican came out as a white blob, and its side camera ends up inside the bird.
🤝 Paid partnership
6
27
5,300
Out of every Sam Altman interview, meet up, and event I’ve seen him host along with the few convos I’ve had with him. This is a horrible portrayal.
The vibes I get from this trailer - “Ahhh AGI bad” “wowww ai can lie let’s unleash it on the public 😈”
Why are we slopping this kind of Hollywood doomer melodrama onto audiences under the guise of a serious portrayal? Can we at least pretend it’s a serious take on OpenAI?
Can we get even a modicum of verisimilitude instead of these skin walker caricatures?
Brett Adcock puts BotQ at roughly 40,000 humanoids a year at full capacity across four lines.
He sees a path to “a million units a year out of a facility” if Figure can crack the next stage of scaling intelligence. That's roughly 25x BotQ's estimated capacity from a single plant.
As well as figure 4 will be the first mass manufactured humanoid out of figure!
Everyone’s benchmarking coding agents on whether they can ship an app.
Volcano’s LavaFlow reads the logs and deployed code, then opens a PR you can review!
🤝 Paid partnership
At Kong we run half a quadrillion AI tokens a year, and I kept seeing agents stuck on clouds built for websites.
So we built a cloud for agents.
Volcano is live today: the self-healing agent cloud. Sandboxes, durable workflows, Postgres, hosting.
1
10
2,391
Hello everyone I’ve been seeing rumors of an Anthropic release from a lot of my friends.
I’m sorry to tell you this will not happen.
Do not expect Opus 5.6 / Fable 5.5 tomorrow.
I am convinced Anthropic has won the race at this point - model 5 is almost done as well!
OpenAI looked at how its own researchers use AI to build better AI. They tracked agent usage, success on delegated tasks, experiment volume, and model benchmark scores. The frontier is “jagged” because agents can do some parts of the job on their own while still struggling badly with others.
The coolest finding in the recent blog however was -46% of delegated 2–4 hour tasks, succeeded without human intervention, up from 23% in March. For 4–8 hour tasks, it went from 13% to 35%. Those are estimated ‘human’ task lengths. Researchers also changed what they handed off, so this isn’t a clean model-to-model benchmark.
What's sad is I spent two days building my own harness to connect Claude Code to Codex, and some startup builds it for me the very next day, and it's much more seamless.
Now I know how it feels to be a startup getting steamrolled by OpenAI.
Please don’t be another Theranos
🤝 Paid partnership
Introducing Helios: measuring 1,000+ protein biomarkers from a drop of blood to help you live younger, longer.
Here’s how it works:
• Self-collect blood at home with a painless, needle-free device
• We measure the proteins carrying signals from cells and organs throughout your body
• Your Helios AI agent explains your results, highlights risks, and offers personalized guidance
Proudly backed by Y Combinator and Formation VC.
Join the waitlist at myhelios.com
4
78
7,089
AI video has gotten good enough that the difference between slop and something you’d actually want to publish is starting to come down to creative direction.
Being able to change one part of a video without regenerating the entire thing is a pretty big deal for that.
🤝 Paid partnership
Frontier AI has solved math. Science. Coding. But it still hasn’t solved taste. Until now.
Introducing Arcade 3.0: the AI video playground for marketing teams.
Type one line, answer two questions, and Arcade does the rest.
Voice, captions, and music included.
Arcade runs its own harness on top of the best models.
It’s seen your actual product AND knows your brand.
The numbers show it: in a blind test, B2B marketers chose Arcade’s video 2x as often as videos from Claude Opus 5.5 and Gemini Argon.
And you can change one thing without redoing the whole video.
Oh, and this video? Made with Arcade.
Comment "ARCADE" in the next 48 hours for 50% off your first 2 months of Arcade Pro.
26
3,853














