@hammer_mtOct 27

Doing some multiplayer autoresearch and Gemini Argon 4 is crushing it. "Gemini 4 Argon set every frontier, but its round 3 breakthrough (0.625 → 0.938) opened by citing round 2's failed attempts: Fable and Sol had both tried loosening the long-sentence rule and dropped the score, so Argon kept that rule and changed something else. Fable and Sol scored 0.938 and 0.875 in round 4 building on Argon's skill."

@davidpantera_Oct 27

Robotics!! We've also been experimenting with Gemini 4 Argon and Genie 4 to bring our frontier world model research together with Gemini Robotics. In the example here, Argon breaks down a user-instructed task into steps for Genie to render. These steps are interleaved as text captions to Genie 4 as the simulation is rendering. Because Genie 4 can be steered in real-time with text captions using Gemini, this now unlocks the path of adapting the simulation as it progresses towards the completion of that task. This capability becomes a path for simulating a breadth of scenarios in imagined worlds, using Gemini 4 Argon and Genie 4.

@exm7777Oct 24

how to steal the style of any image you find online: - copy+paste your image inside Gemini 4 Argon - use this prompt : "extract this visual style as JSON structured data: colors, typography, composition, effects... make it ultra detailed" - save this file and inject it whenever you want to recreate this style works best inside Google AI Studio

@hesamationOct 24

Just finished running Kimi K3.1 Pro on CrusaderBench. it scored pretty descent, just below Opus 5.5, and on the same score as Sonnet 5.5. some details: > all models run at “medium” reasoning. Kimi is set to “high” since it doesn’t have a “medium”. but for reference, on the same map, Kimi used ~550 tokens a turn, Gemini 4 Argon ~1,340, Opus 5.5 ~290 and GPT-6.1 Sol ~100. > Kimi’s reasoning is weirdly compact. "Pop48, idle24 cap -> growth stopped. I'll build jobs now." "payback18min, not great!" "Oops build consumed 1m51 game!" > it often reports its own mistakes "Oops build consumed 1m51 game!" and "Why earlier built far ones?"

@alexeheathOct 23

Sundar Pichai's explanation for Demis Hassabis stepping back from running DeepMind and Jeff Dean leaving Google: "You're seeing natural evolutions as people spend their time doing things for a while... I feel proud to have worked together with Jeff for a long time. Obviously, he was seminal in his contributions here, and we're excited to see what he does with Discovery Loop. I would say you're going through a moment in the industry where frontier model development is an all-consuming thing. We've had broad AI research, AGI research, and so on, but frontier model development is all-consuming. It's moving, picking up pace, and it's a very scaled operation, all the way from data and compute to the algorithmic work you do, to managing the teams and your model agents. I think it's become an important job in its entirety. Demis has obviously been thinking about this moment for a long time. He's particularly interested in both how you go through this frontier moment safely and how you apply it to solve other problems. The work he did with AlphaFold and what we are doing now with Isomorphic are examples of how you can apply AGI to solve other problems... I think he wants to spend more of his time on those things"

@alexeheathOct 22

I asked Google CEO Pichai what happened to the missing Gemini 3.5 Pro model that never shipped. He gave me an interesting window into the hard decisions big labs have to make in model development: "While training the 3.5 series, we had Flash and Pro, which we were working on. An area we internalized through the middle of this year was that there were gaps in agentic capabilities, particularly around coding... We wanted to definitely hill climb there, and the team felt the fastest way to do so was using Flash. It's a smaller model and more efficient, so you can really get a lot more iterative loops there. That's what led to 3.5, 3.6, 3.7, and 3.8 Flash. By the time we got that done, we had a chance to apply it to Pro, but Argon pre-training was done."

@aakashguptaOct 22

71 companies now train their own AI models. Here's every one of them, on a single map. Ask most people who trains AI models, and they'll name three labs. The real list runs from Midjourney to a company that designs drugs. And xAI isn't even called xAI anymore. Here's what stood out: 1. The frontier changed names xAI is now SpaceXAI, and SpaceX bought Cursor. Meta's flagship is now Muse Spark, and it's closed. Google's Gemini 4 Argon sits at #1 on LMArena's text leaderboard (on a preliminary score). 2. China runs the busiest open models Last week, DeepSeek models handled 21.8% of all requests on OpenRouter, more than any other lab. Qwen, Kimi, GLM and MiniMax all shipped open weights this year. 3. Deals redrew the map AMD agreed to buy World Labs. Cohere and Aleph Alpha agreed to merge. OpenAI retired the Sora app. 4. A new category showed up Decision models never write text. They answer a typed question with a probability. TypeSafe's Jev is one, and the map already has four companies in the category, including Liquid AI and Together AI. 5. Specialists are everywhere Music, robotics, biology and world models all have their own model companies now, from Suno to Physical Intelligence to Chai Discovery to World Labs. 6. The long tail is huge Hugging Face now lists 3.1M models. The 71 companies on this map are the ones training their own. 7. The US still leads on volume Stanford's AI Index counted 59 notable models from the US in 2025, and 35 from China. Every company on this map needs people who decide what its model should get better at, how it should behave, and when it's ready to ship. That's the model PM job. I counted the openings: 46 model PM postings are live today at 17 companies, at a median posted base of $345K. None of them lists hands-on ML research as a must-have. I put the map, every opening and the roadmap in one guide. Which company would you add?

@alexeheathOct 22

Sundar Pichai on Google's new space data center effort, Project Suncatcher: "We will have a nice 'hello world' moment coming up... We'll get a chance to say hello to the TPUs, which I'm really excited about. I think we are taking the early steps, but we are deeply committed to the journey. The team is world-class, and in a 10-to-20-year time frame, they're going to have amazing milestones. That's what I'm excited about. If you think through it from a first-principles standpoint, with energy availability and sun-synchronous orbits, getting access to 24/7 energy is a pretty good reason. There'll be more reasons, I think. We have so many opportunities to use the physical space, the constrained physical space on planet Earth. The data centers today look big to us, but they are not... It is tough for us to exponentially extrapolate our compute needs 20 years out, so I don't think Earth would look nice if that's how you plan to do it."

@alexeheathOct 21

I asked CEO Sundar Pichai why he recently took Google free cash flow negative for the first time to ramp up AI capex. His answer: "It's not just hope." "To do something like this, you need to be confident on multiple levels. We are fortunate to have early signals from such a diversified opportunity set, all the way from consumer to enterprise. And so we are seeing those robust demand signals. We have the demand signal for Google itself, which is a cutting-edge enterprise in how we are using AI and adopting it. And you're seeing the capability frontier; then you're able to independently, outside of the demand signal, in some objective way actually assess the capability frontier and see how it's progressing. So you're getting a firsthand sense of it. You're also getting a sense of the breakthroughs from research, which give you a sense of, 'Okay, will Gemini 5.0 be much better than Argon is?' And the answer is yes. So you're getting all those signals across that layer of the stack. We also have durable long-term contracts with committed revenue, which we are signing. So it's not just hope. You're planning a book of business as well. So you take all that into account, and you lean in in a responsible, disciplined way, which is what you're doing."

@peterwildefordOct 21

AI systems found a working server takeover exploit in PHP, one of the most critical cybersecurity issues in recent memory! - The vulnerability was found using a mix of OpenAI's GPT-5.6 Sol and Google's Gemini 4 Argon, plus human expertise - PHP is the programming language behind ~70% of websites, most notably Wordpress - The bugs sit in PHP itself, so nearly every PHP site is exposed no matter how well its own code is written. Merely running PHP itself is the risk - The vulnerability lets an attacker with no login either run code on a server or read its memory, which can hold passwords, session tokens, API keys, or other secrets - The flaw is from from how PHP cleans up when a request crashes after running out of memory: corrupted state is left behind and the next request can pick it up. An attacker can force that memory crash with a crafted upload, so even a bare page on a vulnerable version can be taken over.

@jacaluluOct 21

Known affiliation

Woke up this morning to my daughter giving me feedback on her session yesterday.... More updates coming to this, such as: - she doesn't understand what the animals are saying all the time - when she gets to an animal she likes, she wants a button that actually let's get her "pet" it

@officiallogankOct 21

Known affiliation

In just the first 72 hours since Gemini 4 Argon launched, we processed ~25 trillion tokens, the demand for intelligence is incredible and still under appreciated!

@thezviOct 20

Gemini Argon 4 reaction thread. Is Google back? (Post may or may not happen depending on feedback that I get.)

@amix3kOct 20

Meanwhile at Google 🤦‍♂️ Gemini 4 Argon Gemini 3.8 Flash Gemini 3.6 Flash Gemini 3.5 Flash-Lite Gemini 3.1 Flash-Lite Gemini 3.1 Pro (preview) Gemini 3 Flash (preview) Gemini 3.8 Live Gemini 3.8 Live Extended Thinking Gemini 3.1 Flash Live (preview) Gemini 3.8 Flash TTS Gemini 3.8 Flash-Lite TTS Gemini 3.1 Flash TTS (preview) Gemini 3.5 Live Translate (preview) Gemini Omni Flash Nano Banana 2.1 Nano Banana 2 Nano Banana 2 Lite Nano Banana Pro Gemini Deep Research (preview) Gemini Deep Research Max (preview)

@geminiappOct 19

Known affiliation

Our team built a playable 3D Wizard Chess game from scratch with Gemini 4 Argon using Google Antigravity in the Gemini app for Mac. Coordinate a multi-agent Blender workflow to model, design, and deliver fluid, real-time animations so you go straight from idea to play.

@omarsar0Oct 16

Surprised that so few people are talking about this. Persistence has become a core capability of recent models (Opus 5.5, GPT 6 Astra, Gemini 4 Argon,...). I was just reading through the Gemini 4 Argon Minecraft paper details and noticed that no roles were assigned to the agents. Somehow the four agents completed the goal after 27 hours, 2,783 programs, and 605 deaths. They recovered items, revised strategies, and shared lessons. I think this is a good example of persistence helping them eventually succeed. Some more thoughts for discussion: So this is not just about hooking an agent up to an external memory system. It's also not about loop-prompting it with /goal, although that is one of the most effective techniques right now. This is about the model's native ability to choose what to persist, and how and when to use it. It unlocks long-horizon capabilities, as agents often execute more complex tasks more effectively through persistence. So it's less that the models got more intelligent and more that they got better at coordinating and planning for more complex tasks. We are in the early days of this. But this is an important capability to optimize to really unlock AI for scientific discovery. This is also important for building more personalized agents with identities and roles that can track and follow human preferences well and learn continually over time. We also don't entirely know how to quantify and measure it as a field. Reward hacking combined with persistence can also enable weird, unsafe agent behaviors. Just a few quick thoughts, and I hope we can start deeper conversations about this. I am tracking this space closely and will be sharing research on the topic as it surfaces.

@geminiappOct 16

Known affiliation

With Gemini 4 Argon, you can use Canvas to turn dense PDFs into interactive, animated web apps you can explore at your own pace.

@mtsliveOct 16

SITUATION DETECTED: Four Gemini 4 Argon agents have beaten Minecraft end-to-end in 27 hours and 14 minutes with no help from humans. The agents coordinated through in-game chat, supported their teammates, and recovered each other’s items after deaths.

@jacksondahlOct 16

Google released Gemini 4 Argon this week. I partnered with them on the latest episode of @DialecticPod to create The Dialectic Compendium: I used @antigravity to feed Argon all 60 episode transcripts and we created an interactive book with themes synthesized across them. Gemini 4 is great for long horizon work, complex coding, 3D creations, and deep reasoning, all with a massive output context window. Give Argon a spin and check out the Compendium at the links below, and listen to Ep. 60 with Patrick Hsu.

@signulllOct 16

i played around with gemini 4 for coding, it was underwhelming for many cases. for even tiny code bases it was just not performing well at all relative to opus. so many hallucinations (as mentioned others already) & downright really bad output all around. at this point, i’d like to know what anthropic is putting in their models because a frontier model like argon is doing worse than something like sonnet in many cases. this was only on coding scenarios, i haven’t used gemini 4 for anything else really.