Gurupreet singh (@_gurupreet) header
Gurupreet singh (@_gurupreet) profile picture

Software Developer @Google, Composing Android Automotive ex Dev-Rel #Android #MaterialDesign #JetpackCompose Opinions are my own. https://teapawt.org/t.co/FbzO31YSpW

Germany
Joined July 2012
Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)Photo by Gurupreet singh (@_gurupreet)
Gurupreet singh retweeted
Antigravity context window is now doubled for Gemini 4 models!
We’re doubling the context window for Argon, Carbon, and Hydrogen in Antigravity. Lots of work to make this happen but you should now see compaction go down by 3.5x per user turn!
69
29
6
1,023
40,685
Gurupreet singh 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 🧵)
171
155
73
1,918
494,665
Gurupreet singh retweeted
Sharing early evals on Gemini 4 Carbon. Strongest model that we've seen so far for long-form research and answer quality, with a new SoTA on the Wide-And-Deep-Research (WANDR) benchmark. Coming soon to all Perplexity Pro and Max users inside Computer, once we have sufficient token capacity. Congrats @GoogleDeepMind!
179
209
70
2,899
249,757
Gurupreet singh retweeted
our team uses gemini quite a bit now whenever we have a tricky architectural problem, complex system design questions, or need to simplify something we ask for a gemini subagent it's pretty good here. i don't see this talked about much
299
94
70
4,437
374,664
Gurupreet singh retweeted
A thread of all the amazing outputs from Gemini 4 Carbon so far!! Can't wait to try it out when it officially comes out.. 1/ Ethereal black hole horizons
21
49
5
777
40,557
Gurupreet singh retweeted
Google's latest model (which is announced but not publicly available yet) just jumped into the lead. Everyone talks about Google dropping the ball. I keep saying over and over again, never count out Google.
Introducing Gemini 4 Carbon, it's a big jump over Argon which we released just 2 weeks ago. Gemini 4 Carbon is SOTA on cyber, long horizon tasks, ranks #1 on Artificial Analysis V5, significantly improves on usability in domains like coding, and is more aligned.
24
4
3
106
8,310
Gurupreet singh retweeted
Genie 4 is pretty impressive. I find the visuals to be very realistic and high-quality. The ability to prompt anything into the world and have it just appear opens up all sorts of possibilities to steer a scene in whatever direction you want. I found it really difficult to control the object on the screen, the monkey, the car, and the bi-plane all sort of ignored my attempts to steer them. There was also some temporal coherence issues where I'd turn around and face a different direction and then return to my original position and things would have changed. The Mustang, for instance, changed directions when I turned around. When it drove off later in the scene, it turned into a taxi... Little things like that change over time. Visually, however, it's really impressive. (IMO)
22
26
4
249
15,753
Gurupreet singh retweeted
We got GTA Waymo before GTA 6 (this is fully interactive real-time video from Genie 4)
107
105
42
1,951
247,183
Gurupreet singh retweeted
Had early access to Genie 4 from @GoogleDeepMind. This is a fully generated, interactive world from a single starting prompt (environment + character) 🤯 Feels like a massive step forward - you can now prompt to change the world and interact with it. With native audio!
152
187
74
2,368
209,175
Gurupreet singh retweeted
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.
1,030
2,636
2,053
23,835
3,550,243
Gurupreet singh retweeted
Gemini 4 Argon is underrated 😱. When auditing the traces, we found that Pi Harness just ends the run whenever it hits a fallback (e.g. for security task) in 20% of the runs. We rerun Argon with Antigravity Harness, and now it beats Astra and sits right behind Opus 5.5. Congrats to the team @GoogleDeepMind and looking forward to the new ckpt.
FrontierCS 2 is our new benchmark for CS research & development. Each task is based on a paper: we delete the contribution from the released code, and ask agents to outdo the original method, as judged by the authors' chosen metrics. Agents don't have access to the paper or web.
59
73
79
1,019
348,887
Gurupreet singh retweeted
Congrats to @GoogleAI's Gemini 4 Argon for taking the top spot of our Humanity's Sixth Sense (HSS) benchmark with a score of 55.4%, edging out GPT-6-Astra at 53.6%. 🥇 HSS tests AI’s ability to reason beyond what’s visible, from hidden objects to predicting what happens next and understanding human behavior. We took a closer look at performance across categories, and that’s where things get interesting. 👀🧵
40
129
41
1,166
162,906
Gurupreet singh retweeted
I have to admit, watched this a few times, especially to the end:) the creativity in the prompting and the model is awesome to see
Argon has this amazing natural understanding of the world. It can do so many things from first principles. Here's a pelican riding a bike, using fourier epicycles. Video made by Argon. Time-synced Lyria 3.5 music prompted by Argon.
316
523
41
9,150
953,877
Gurupreet singh retweeted
HOLY SHIT. google really cooked it with argon this time. it was the fastest of all (same prompt).
43
55
7
2,305
308,678
Gurupreet singh retweeted
Gemini 4 Argon 和 Astra 的对比测试 完全一样的提示词 Astra 耗时 25 分钟,成本 11 美元 Argon 耗时 3 分钟,成本 0.5 美元 速度差了 8 倍多,成本差了整整 22 倍, Argon 完全碾压了 Astra。 如果不细看,我甚至以为右侧是一张图片。
77
13
7
411
120,686
This is crazy! One of the best demos of Gemini 4 Argon, showing what it's capable of. Gemini 4 Argon taught a robot to play cornhole just by watching a video.
11
10
2
277
10,514
Gurupreet singh retweeted
Wow Google has really impressed me, Gemini 4 I was expecting to be a benchmaxxed trash model but it actually competes with the current frontier like 6.1 sol, opus 5.5 etc… Good job on your hard work Google!
29
35
1
1,065
43,109
Opus 5.5 vs Gemini 4 Argon Same prompt. Same task: build a cinematic 3D animation explaining how a Tesla-style electric motor works. Which one do you think did better?
148
108
27
2,136
196,188
Gurupreet singh retweeted
Gemini Argon 4 is a frontend beast. Do you think Opus 5.5 or Argon 4 won here? Sonnet 5.5 vs Argon 4 is up next.
24
4
63
3,446
Gurupreet singh retweeted
omg look at this!! it seems Gemini 4 Argon is a demon in 3D rendering. this video demo looks straight out of unreal engine.
24
15
3
277
11,780