Pinned Tweet
Sundar Pichai: In three years, today's AI will look as primitive as a flip phone. The progress we're about to see will surprise people.”
“We are in that rare moment in technology. The curve is so steep.”
“People will look back three years from now and say, ‘You mean you couldn't do those things?’”
“We are all laying the foundation for agents, but agents will be very common.”
“People will be relying on them to get things done in a more seamless way.”
“Any content can go from any modality to any modality. We've shown hints of it with Nano Banana, Veo, and Omni.”
“It doesn't matter how the content was originally created. People can consume it in many different forms.”
Sonnet 5.5 > Gemini 4 Hydrogen
I gave both models the same prompt at extra high reasoning and the two ships came out nothing alike
> sonnet 5.5 took ~32 mins and cost ~$2.60 in claude code
> hydrogen took ~15 mins and cost ~$1.2 via Cursor
Hydrogen finished in half the time, but the first pass needed a V2 before I got an output I could use.
Sonnet 5.5 is still one of the most creative models for 3D coding.
14
44
1,627
Gemini 4 Hydrogen vs Opus 5.5 (Tandem Ride SVG)
same prompt: create an SVG of a girl and her cat on joined bikes, golden hour, ghibli style
• opus 5.5 in claude code: ~1 hr, ~$18 api cost
• gemini 4 hydrogen in cursor: ~15 mins, ~$0.9 api cost
opus took its time with the detailing (night mode, music) while hydrogen got the whole scene done faster for less than a dollar.
which model did it better?
Gemini 4 Hydrogen (Flash) is Live!
- users are already reporting access through Vertex AI, Antigravity and AIHubMix, including some Pro users.
Google hasn’t made the official announcement yet, but we should expect one anytime soon.
the Hydrogen bomb is finally here, what are we testing first :)
5
29
1,607
Jensen Huang at NVIDIA’s board meeting today
$NVDA - BREAKING: NVIDIA CEO JENSEN HUANG RESIGNS EFFECTIVE IMMEDIATELY
Nvidia announces a major leadership shake-up:
🔸 Jensen Huang resigns as CEO, president and board member, effective immediately.
🔸 Colette Kress, Nvidia's CFO, appointed interim CEO.
🔸 Harvey C. Jones named board chair.
🔸 Search for permanent CEO begins, considering internal and external candidates.
Huang's departure ends his 33-year tenure leading the AI chip giant he co-founded.
1
38
7,806
Dario and the other six founders plan to sell roughly $250M in shares each in the IPO
Sam Altman right now:
Our initial public offering: ipo.anthropic.com
Legal disclaimer: Anthropic has filed a registration statement (including a prospectus) with the SEC for the offering to which this communication relates. Before you invest, you should read the prospectus in that registration statement (available at: sec.gov/Archives/edgar/data/…) and other documents Anthropic has filed with the SEC for more complete information about Anthropic and this offering. You may get these documents for free by visiting EDGAR on the SEC Web site at sec.gov. Alternatively, copies of the preliminary prospectus may be obtained from: Goldman Sachs & Co. LLC, Attention: Prospectus Department, 200 West Street, New York, NY 10282, by telephone at (866) 471-2526, or by email at [email protected]; or Morgan Stanley & Co. LLC, Attention: Prospectus Department, 180 Varick Street, 2nd Floor, New York, NY 10014.
The securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post is not an offer to sell, or a solicitation of an offer to buy, the securities in any jurisdiction where the offer or sale is not permitted.
4
26
1,778
Gemini 4 Carbon vs Opus 5.5
tested both with the same prompt [Voxel Floating Kingdom]
• Opus 5.5: high reasoning, 1h 15m, ~$12
• Gemini 4 Carbon: high reasoning, 1h 15m, ~$0.95
at that price difference, Carbon is looking like the better deal. but which model actually did better here?
Happy Gemini 4 Hydrogen Day?
Logan just posted “Gemini” again, exactly what he posted before Carbon dropped yesterday. Hydrogen could be next.
here’s what we should expect based on early checkpoints tested on Arena as “gemini-3.8-flash”:
> fast generation, especially for coding and big jump over 3.8
> strong 3D coding with lots of detail and creativity
> much cheaper than the larger Gemini 4 models, as expected of a Flash model.
if Google can deliver this level of quality at Flash speeds and pricing, it's going to be a seriously competitive model.
are you excited for the Hydrogen bomb?
11
58
1,897
Gemini 4 Carbon (Voxel Floating Kingdom)
- high reasoning in Cursor, ~ 1h 15m ~ $0.95 in API costs
over 3 million voxels, this much detail and creativity, all for less than a dollar. (est. 11.7M tokens)
easily the cheapest SOTA-level model i’ve tested so far. the price-to-quality ratio is impressive.
Google cooked insanely hard with this model
Gemini 4 Carbon is now available.
It’s our strongest model, built to dig deeper, handle long-horizon tasks, and multi-step reasoning.
If Argon is great for complex tasks, Carbon is the model you bring in for your hardest problems.
11
38
1,853
POV: OpenAI in the AI race after Gemini 4 Carbon and Fable 5.2 dropped
We're turning our lessons from defending OpenAI into practical tools for our customers. On Thursday, join us at Cyber at the Frontier, a live keynote, for our latest cyber announcements.
Working with our frontier models, we've identified and resolved important vulnerabilities in our own code and infrastructure. That experience is informing the tools and playbooks we're building to help other security teams strengthen their defenses.
@gdb shares what we've learned and why security teams should keep putting the latest models to work as capabilities improve.
Tune in on November 12 at 10 a.m. PT: openai.com/cyber-at-the-fron…
Fable 5.2 vs Gemini 4 Hydrogen (Pelican SVG Test)
same prompt: create an svg of a pelican riding a bicycle across five animated seasons.
• Fable 5.2 (high, Claude Code): ~25 mins, ~$25
• Gemini 4 Hydrogen (Arena): ~5 mins, cost unknown.
Hydrogen held up surprisingly well against Fable and could be 10x cheaper than Fable when it launches.
Fable 5.2 did a really impressive job overall but is the extra detailing worth the cost?
Grok 4.8 is finally coming this week
Elon says the 2.5T model has finished pre-training and is now in RL and it’s a major jump over 4.7
xAI also rebuilt the RL and inference systems from scratch in C/C++, which explains the long delay
the model is set to begin rolling out to Grok Bot users later this week
one question: will it be on par with Opus 5.5 when it drops?
Replying to @cradrrat
We have dumped JAX/SGLang and rewritten the entire RL and inference stack from scratch in C/C++, which took some extra time.
Training (aka pre-training) is done. Lot of cleanup was needed on the SFT data to get that mostly right. Now, 4.8 is in RL and the evals are starting to climb. We will do a soft rollout to Bot users later this week and see how it goes.
Grok 4.8 will bear no resemblance to 4.7 (thankfully). It is not yet incredible, but, this time, the foundations are built right, so I expect it to evolve to be something special.
the Anthropic effect is already kicking in
less than 24 hours after Fable 5.2 launched, Gemini team is already back to vague posting.
Carbon has already appeared in Google Cloud’s billing catalog, and new Hydrogen checkpoints have been testing on Arena for days now.
the timing is interesting. we might see wider Carbon access or Gemini 4 Flash sooner than expected.
Fable 5.2 (3D Sailing Ship)
- medium reasoning on claude code, est. ~38 minutes ~$17
Anthropic cooked with this model. it built this in Three.js, no Blender or external 3D assets were used
we’re getting to the point where 3D scenes built entirely with code are starting to look like actual renders.
OpenAI really has to step up now and drop 6.2 Astra
the quality might look a bit poor in the video.
check it out here claude.ai/artifact/7ge282etT…
5
447
Dario after ending Gemini 4 Carbon’s five seconds at the top with Fable 5.5 and making it available to Pro users

















