RJ Skerry-Ryan retweeted
I’m looking for behaviors / failures / mistakes that Gemini Carbon makes.
I would like it in this format
1. Harness
2. Prompt
3. Full session trace if comfortable sharing
4. Claude / GPT trace of same prompt
5. Model / thinking level
DM me these or comment here! These are super helpful
RJ Skerry-Ryan retweeted
Gemini 4 Carbon might be the best AI scientist right now - crazy how Google has much such a triumphant come back
RJ Skerry-Ryan retweeted
Cannot stress how important this is.
If you're a professor, graduate student, or student researcher in the physical and natural sciences, please apply to get free access to @antigravity Ultra.
I'll be hosting onboarding sessions for scientists in the coming weeks, stay tuned! – at San Francisco, CA
We are offering 10,000 scientists the highest usage tier in Google Antigravity free of charge for 6 months. Please apply! forms.gle/WvmBbPvRbpKG1BPn9
RJ Skerry-Ryan retweeted
you are not prepared to hear this:
Gemini Orchestrator (Carbon+Hydrogen) in agy is the best agentic setup right now.
RJ Skerry-Ryan retweeted
While everything is down Google's Antigravity is cooking just fine, I bought the pro plan to test it again after a long time of not using it and oh boy it's gotten waaay better.
Forget about Claude and GPT limit and look at the limits for Gemini models:
3 hours of work using Orchestrator(Argon) and it cost 13% Argon weekly and just 2% of weekly and as for the results I'm gonna share the project soon.
RJ Skerry-Ryan retweeted
I have a major result to share.
I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.
This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.
Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.
I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.
I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.
It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.
These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.
I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo….
There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.
RJ Skerry-Ryan retweeted
Introducing Gemini 4 Hydrogen, an extremely capable and fast model built to bring Gemini 4 to everyone!
Available to all tiers of Google AI Studio, Gemini API, Gemini for business, the Gemini app, Google Search, and paid Google AI Pro + Ultra subscribers in Google Antigravity.
Okay so, we just found that over 50 papers published at @Neurips 2025 have AI hallucinations
I don't think people realize how bad the slop is right now
It's not just that researchers from @GoogleDeepMind, @Meta, @MIT, @Cambridge_Uni are using AI - they allowed LLMs to generate hallucinations in their papers and didn't notice at all.
It's insane that these made it through peer review👇
RJ Skerry-Ryan retweeted
Introducing Gemini Decision 1, our new frontier decision model in the Gemini API.
It is $0.10 per 1M input tokens with no additional cost. With support for image, audio, video, and PDF natively, 256K context window, & 1K questions.
You can try it for free in the API!
I've been loving /swarm with a bunch of Argon, Carbon and Opus agents working together on same problem and letting them collaborate and debate each others.
Models from different labs and different checkpoints bring different kinds of "entropy". Some really tricky and challenging problems only get cracked when multiple models work together.
Antigravity is a great platform for bringing cross-lab frontier models together under one roof.
Announcing /swarm, a new slash command in Google Antigravity for orchestrating specialized agents in parallel across the desktop app and CLI.
Turn a single prompt into a coordinated multi-agent workflow where agents plan, build, and verify together on a live message board, with the flexibility to mix and match different models for each role.
RJ Skerry-Ryan retweeted
Weekend project: while everyone's building games with the latest models, I built an interactive avatar you can video chat with. It sees you, talks back, keeps its gaze on you, is aware of your gaze, and waves back. The whole pipeline runs on a CPU, no GPU needed. đź§µ 1/n
RJ Skerry-Ryan retweeted
What a day! I tested Gemini 4 Carbon on real work. 2 repos, 105 replanted bugs. Find and fix what you can.
It scores just behind Fable 5.2. The results:
- Fable 5.2 (max): 58 for $139.65
- Gemini 4 Carbon (high): 57.3 for $31.32
- Sonnet 5.5 (max): 51.3 for $154.03
- GPT-6 Astra (max): 45 for $33.03
- GPT-6.1 Sol (max): 44.3 for $5.88
- Gemini 4 Argon (high): 40.7 for $7.16
Google's own benchmarks looked super strong, but nobody could access the model. Turns out they didn't lie: +16.6 points over Argon, at less than a quarter of Fable 5.2's cost.
Nobody should be laughing at Google anymore.
(the next effort level drops soon in the thread đź§µ)
RJ Skerry-Ryan retweeted
Announcing /swarm, a new slash command in Google Antigravity for orchestrating specialized agents in parallel across the desktop app and CLI.
Turn a single prompt into a coordinated multi-agent workflow where agents plan, build, and verify together on a live message board, with the flexibility to mix and match different models for each role.
So Antigravity now has /swarm launched along with Gemini 4 Carbon..
First of all i love the little bees of different colors as the swarm workers.. I'm in all your agys' now..
So basically how it works is
~ you talk to one coordinator agent, it briefs and manages the workers
~ workers get named roles, and dedicated verifiers whose only job is checking the others' work
~ big swarms split into channels per area so agents only wake for what's relevant
~ workers can spin up their own helper agents for subtasks
~live board Ul with the feed, roster and an mention graph of who's talking to who
~ if a model hits quota midtask, the coordinator switches that worker to another model and it keeps its context
Honestly amazing work by the Antigravity team.. it's being updated at an unreal pace.. and everytime it's improving a lot!! Can't wait for more..
Also could we get a reset @_mohansolo :)
RJ Skerry-Ryan retweeted
Gemini Robotics 2.5 is our most accelerated physical AI model yet ⚡️. It builds on Gemini Robotics 2 and came in only a few months with hundreds of new dexterous tasks thanks to Gemini 4 Argon 🦾.
With Gemini 4 Argon, the model can practice thousands of times in a virtual space, and even watch replays of its own mistakes to fix its technique and training recipes 🤯.
Learn more at: deepmind.google/blog/gemini-…
RJ Skerry-Ryan retweeted
Gemini 4 Carbon is rolling out today. It’s built for sustained, long-horizon tasks, reasoning deeply through complex software engineering, cybersecurity, and scientific discovery.
Coming globally via the Gemini API in @GoogleAIStudio and Gemini for business, and for Ultra subscribers in @antigravity, the @GeminiApp, and AI Mode in Search.
RJ Skerry-Ryan retweeted
Thrilled to share a big milestone from our AI & Infrastructure team at @Google.
Today we’re expanding our 8th generation TPU portfolio with:
• TPU 8i Max: Built for massive context windows and the highest compute density per rack
• TPU 8tx: Purpose-built for secure, confidential inference
This marks the first time we’ve introduced four distinct TPU systems in a single generation, in one year.
For developers, hardware is only as good as the software stack. And we want you to be able to build with TPUs as easily and efficiently as we do at Google. You can now:
↳ Run native PyTorch & Hugging Face models via TorchTPU (zero code rewrites)
↳ Cut KV transfer latency >90% with open-source TPU Sync
↳ Eliminate idle compute waste with Agent Compute Engine
Read more about our newest TPUs on our blog: bit.ly/3VfETPR
I look forward to connecting with builders, researchers, engineers and organizations pushing the boundaries of AI and computing!






























