Doriana (@astierun) header
Doriana (@astierun) profile picture

Founder @ Studio.lab Product Lead & Co-founder @ Anicry Check this — https://xitter.twitterapi.xyz/t.co/71lDuf7pDn

Ukraine
Joined June 2023
Photo by Doriana (@astierun)Photo by Doriana (@astierun)Photo by Doriana (@astierun)Photo by Doriana (@astierun)Photo by Doriana (@astierun)Photo by Doriana (@astierun)
I finally shipped Studio.lab 2.0 ✨ A free, open-source extension for Google AI Studio: redesigned chat, draft recovery, Markdown tools and smoother long chats. This project grew far beyond what I imagined. Would love your thoughts, @googleaidevs!
1
2
267
Carbon in AI Mode is Ultra-only and the country list runs 122 deep with zero EU states in it. Strongest model Google has, gated by a $100 plan and a map.
Shipping Gemini 4 Carbon in Search today for Google AI Ultra subs! With deeper reasoning & better instruction following, you’ll get more helpful responses for your toughest questions in AI Mode. I’ve loved using this internally – excited for more people to try it out.
4
Google shipped a model that never writes a word. 1,000 yes/no questions, one pass, probabilities back. Agents have been paying full price to generate paragraphs just to fish a yes out of them. That tax is gone.
introducing the Gemini Decisions API powered by Gemini Decision 1, our new multimodal decision model get fast, calibrated decisions on up to 1000 questions at once across text, images, audio, video, and PDFs try it today in AI Studio and via the Gemini API ai.studio
9
one-shot builds vol. 4, all Carbon @marcthecreatorr: floating voxel kingdom, 3M voxels, $0.95, zero fixes @leppppp1: a mountain town in raw WebGL, 4800 lines, no 3D library @fengxi5678: a 1994 game ported from its binary in one run video: @marcthecreatorr
15
DeepMind put Apollo humanoids on a live Mercedes line and they packed car parts at 96%. The robot demo era just got a factory floor.
Gemini Robotics 2.5 is here. 🤖 Our latest physical AI is able to learn hundreds of everyday skills - from folding laundry and serving pastries to autonomously sorting car parts on the factory floor. Here’s a snapshot of its capabilities 🧵
13
Anthropic put a price on it: $70 to $78 a share, 825 million of them. Up to $64 billion, the biggest AI listing ever attempted. And Google is buying $10 billion of its rival. That last part is the whole story.
Anthropic has launched the roadshow for its proposed initial public offering: anthropic.com/news/roadshow-… Legal disclaimer: A registration statement relating to these securities has been filed with the SEC but has not yet become effective. These securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post does not constitute an offer to sell or the solicitation of an offer to buy these securities, nor shall there be any sale of these securities in any state or jurisdiction in which such offer, solicitation or sale would be unlawful prior to registration or qualification under the securities laws of any such state or jurisdiction.
8
Carbon is actually out now. API + AI Studio for paying devs, Ultra in the app and Antigravity. $2/$10 until Nov 30, then it doubles. same clock Argon had.
Our strongest frontier model, Gemini 4 Carbon, is now available. Built for deeply complex, long-horizon workflows that push a frontier model’s intelligence, Carbon delivers frontier performance in agentic workflows across complex knowledge work, software engineering and web development, cybersecurity, and scientific reasoning. While Gemini 4 Argon remains the ideal model for the majority of software engineering and knowledge work, Carbon steps in for your hardest tasks that push a model to its limits — from complex, multi-step debugging to open-ended scientific research. Available starting today: — Consumers: use it in the @GeminiApp and AI Mode in Search for Google AI Ultra subscribers — Developers and Enterprises: build via the API in @GoogleAIStudio and @AndroidStudio, Google @Antigravity for Google AI Ultra subscribers, and Gemini for business — Scientists and Researchers: apply for a chance to get 6 months free via @Antigravity — Cybersecurity Defenders: apply to join the Fairwind Program for access to Carbon for cyber Watch how Carbon brings reasoning and coding together, turning avian research into an interactive 3D simulation of a songbird’s vocal organ that synthesizes the sounds of six bird species live.
1
23
OpenAI built a kill switch for its own research runs. 35 seconds from decision to dead output. Last time a run went sideways it took 2.5 hours by hand. Internal infra only, not an API feature. The 2.5 hours is the part that sticks.
We built Control Room, a central system to track model workloads across our research infrastructure and stop them quickly when safety or security concerns arise. It connects every research inference workload, including agentic sampling for model training and evaluation, data processing, and more. This works together with our misalignment monitoring systems, and halts model output within 35 seconds of a decision to intervene to stop or pause a job. Control Room is just one part of our holistic approach for safe frontier model training and deployment. openai.com/index/control-roo…
19
Logan just posted one word: Gemini. Last time he did that, Argon dropped the same day.
24
Antigravity just shipped a layout for people who write and research instead of living in a terminal. Same agent, same tools, no git clutter in your face. About time the interface caught up with who actually uses it.
Today, we are launching a new Knowledge Work interface layout tuned for research, writing, and artifacts: same agent harness, plugins, and MCP tools, without the terminal and git clutter. Select your layout in the General tab of Settings. Our goal with Antigravity is to help the world deliver value through products and research. Doing that always starts with building the logic, which is why software development is at the core of the product. But turning that work into real outcomes takes every role working together with a shared language and understanding.
26
Meta's smallest model just showed up on Google Cloud. Ember does Haiku-level work at about a third of the cost per task. $0.10 in, $0.50 out. The frontier stopped being the interesting price tier a while ago.
Muse Ember, Muse Spark 1.3, and Muse Wonder are now available on Google Cloud.
16
Fable 5.2 is out. $8/$40, outputs up to 30% faster than 5.1. Mythos stays behind the verification programs though, so the model everyone actually wants to try is still the one you can't touch.
Claude Fable 5.2 and Claude Mythos 5.2 are available now:
21
If you're on Google AI Plus, check your inbox. From November 13 you lose Pro models in the Gemini app and AI Mode. Flash and Flash-Lite stay. Pro moves to the $19.99 plan, and your old Pro chats carry on with a different model 👇👇👇
1
31
One-shot builds, vol. 3: @HavisT83099 — frog, tongue swing. Argon 2 min, $0.12. Opus 14 min, $2.30. @Rakel076 — sailboat race. Argon won, $0.78. Astra lost, $3.58. @Lili4fhg — four racing sheep. $0.30 took the bush. $2.35 finished. The bill is the review. video: @Lili4fhg
2
1
100
yeah the iteration tax is the whole story. do the cleaner Opus results ever win you back, or does Argon take it the moment you need ten tries?
30
If your code still calls gemini-omni-1.1-flash, December 3 flips it to 1.2 on its own. No error, no warning. Same string, different model answering, cheaper per second. Retest your prompts before the reroute does it for you 👇👇👇
2
1
47
Argon doubles in price on December 1. $2/$10 per million tokens becomes $4/$20. 3.8 Flash follows a month later. If your pipeline runs on Argon, December is when it stops being cheap 👇👇👇
1
13
Gemini 3.8 Flash scores 74% on DeepSWE. Change 17 lines in the task and it drops to 53%. Same model, same test family. The score only holds while the spec stays exactly as written. If you pick models by leaderboard, you're picking by how well they memorized the test. 👇👇👇
1
17
One-shot builds, vol. 2: @camperaigg — lightning, Fable 5.5, part 14. @karanC_12 — voxel tortoise, one shot. The other model gave him bookshelves. @akshay81844 — Kharcha Bolo, offline Hinglish expenses. One run, 31 min, $0.27. The bill is the review. video: @camperaigg
2
109
Anthropic's '20x more usage' pitch just hit a wall in court. A judge said it could plausibly mislead, because 20x only holds inside a five-hour session. Over a full week it's closer to 6x. The fine print was two clicks away. 👇👇👇
1
16
AI Studio has a rate limit page that shows the ceiling. Cool. It still doesn't show burn rate. So you still find out you're out of quota the same way as before, by hitting the wall.
1
14