My PhD thesis "Deep Learning with Graph-Structured Representations" is now available for download: hdl.handle.net/11245.1/1b63b… -- It covers a range of emerging topics in Deep Learning: from graph neural nets (and graph convolutions) to structure discovery (objects, relations, events)
other than the fact that everything takes 20 minutes, gemini 4 carbon is now my favorite model. i use it for anything i want done thoroughly and correctly. it's brilliant, legible, and fixes its mistakes quickly and honestly if asked. it's like if astra max was normal
Meet Gemini 4 Hydrogen: built for rapid coding, multimodal synthesis, and everyday workflows.
It finishes tasks 2.5x faster than 3.8 Flash and surpasses Argon on the @ArtificialAnlys Intelligence Index.
Gemini 4 Carbon is now available for users, developers and enterprises!
We are heavily using Carbon internally at Google for hardest long horizon tasks, our teams have used it to help mathematically prove the core cryptographic protocol of Signal — and it’s topping independent benchmarks like CursorBench 4.0, Perplexity’s WANDR, Zapier’s AutomationBench, and more.
The team also had fun using Carbon to analyze images and technical drawings from humanity’s first crewed lunar lander, the Apollo Lunar Module. Not a single flown component ever returned to Earth. It recreated the 3D asset from those historic documents and generated this walkthrough video.
Read more: blog.google/innovation-and-a…
Meet Gemini Omni 1.2 Flash ⚡️
We’ve made the model more cost-effective, bumped the video generation and editing limit up to 15 seconds, and made updates to how scenes hold together over time. We've also given character speech an upgrade, reduced visual glitches like jump cuts, and made sure the model stays even truer to any reference assets you upload.
So… what does all of that actually mean for you?
It gives you more room to tell your story. With up to 15 seconds to work with, you have more flexibility to create what you need — videos look more cinematic and smooth, offering stronger consistency that keeps objects anchored from start to finish. And when characters speak, the dialogue sounds more fluid and natural.
Creators and everyday users can start directing scenes and multi-turn edits in @FlowbyGoogle, generating 15-second clips at no additional cost in Google Vids, or creating 15-second videos in the @GeminiApp. Plus, for developers and businesses building with the API, we've dropped the price by 30% from the previous generation, making it available at just $0.07 per second for 720p video.
Watch the model in action as it turns a single style frame into a 4-shot, 15-second macro science explainer of a living beehive ↓
The Ocean Cleanup is now $40M/year short of what’s needed to stop up to 1/3rd of all plastic flowing into the oceans by 2030.
That’s less than 1% of what America spends on Halloween decorations each year.
We’ve released AA-Omniscience v1.1, updating our knowledge and reasoning benchmark with a refreshed public set, removal of saturated questions, and ambiguity fixes
AA-Omniscience is our proprietary benchmark measuring model knowledge and hallucinations, including whether models admit when they don’t know an answer. Our primary metric, Net Score, rewards correct answers, penalizes incorrect answers, and treats abstentions neutrally.
AA-Omniscience was updated to v1.1 as part of the Artificial Analysis Intelligence Index v5 release.
Key changes in AA-Omniscience v1.1:
➤ Updated 3,000-question dataset: Down from 6,000 questions, with previously public and saturated questions removed, errors and ambiguities corrected, and new questions added
➤ Newly calibrated 300-question public set: AA-Omniscience v1.1 has a new public set on @huggingface calibrated to match the difficulty of the private set. At release, public and private scores have a 0.99 Pearson correlation and 0.99 Spearman rank correlation for AA-Omniscience Net Score, with a median difference of 0.0 points. Accuracy and hallucination rate also have 0.99 correlations, with median differences of -0.07 and -0.29 percentage points, respectively
➤ Slightly updated domain distribution: Questions span the same 6 domains and 42 topics as version 1, with a revised distribution: Science, Engineering and Mathematics (600), Software Engineering (600), Finance (450), Health (450), Humanities and Social Sciences (450), and Law (450)
Key results for AA-Omniscience v1.1:
➤ Rank order largely preserved: AA-Omniscience v1.1 has a 0.98 Spearman rank correlation with v1. The recently released Gemini 4 Carbon (high) leads with a Net Score of 54, followed by Claude Opus 5.5 (48, max with fallback) and Claude Fable 5.1 (46, max with fallback). Muse Spark Open (xhigh) is the leading open weights model, 28 points behind Gemini 4 Carbon
➤ Scores average ~2 points lower on the refreshed set: Models score 1.9 points lower on Net Score on average compared to v1, with a median decrease of 0.6 points. Some models see much larger declines, particularly MiniMax M3.1 Flash, down 22 points (16 to -6), and MiMo V2.6 Pro, down 13 points (8 to -5)
➤ Slightly lower accuracy: AA-Omniscience Accuracy measures the percentage of questions the model answers correctly. Models answer a median of 1.5 percentage points fewer questions correctly on average compared to v1. Claude Fable 5.1 (max with fallback) has the highest accuracy at 68%, followed by Claude Opus 5.5 (max with fallback) at 66%
➤ Hallucination rates decrease: AA-Hallucination measures how often models choose to answer when they do not know the answer. The median Hallucination Rate falls 1.4 percentage points, with Gemini 4 Argon (high) recording the lowest hallucination rate at 4%
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model.
Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching.
We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards.
As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
This is Genie 4, our most realistic and interactive world model yet. 🌐
Explore dynamic environments in real time to see how the scene responds, sounds, and how the people in it move, react, and speak.
Hi all, I have a new result I'm excited to share. It turns out that *all existing implementations* of softmax and cross-entropy (that's JAX, PyTorch, textbooks, etc) have a subtle bug --- two bugs, actually. A simple rewrite fixes both, and eliminates "slingshots" when grokking:
We added Gemini 4 Argon to our AutoResearchExam Pareto curve. It is 2-3 times cheaper compared to the other frontier models for almost the same ML engineering performance. Congratulations to the GDM team.
As a German, it is a tragedy: in a parallel universe this would be Germany’s moment, as it requires so much that Germany has traditionally excelled at: industrial capacity, engineering, world-leading physics, design and manufacturing excellence. That was exactly what ‘made in Germany,’ stood for.
We’re rolling out Gemini 4 Argon in Antigravity to paid Google AI Pro subscribers today. Demand has been incredible and we've scaled capacity deliberately to get here. We’ve rolled out a separate weekly-only Argon quota for Pro on top of the existing Gemini bucket for 3.8 Flash so you have the best of Gemini for all your tasks.