81 Mac apps are on sale for Black Friday 🖤
Soundbar and Levers: 76% off
Vitals: 73% off
MightyBench: 62% off
ProtectMyMac: 60% off
76 more are waiting in one place 👇
https://t.co/vAGXWCVmWo
🖤 ZenVoice’s Black Friday sale is live on MacBase!
Get 40% off with code BLACKFRIDAY.
Dictate emails, notes, or messages in any Mac app. ZenVoice removes filler words and fixes punctuation on your Mac. Pay once, keep it.
Thanks @attacomsian for featuring us!
https://t.co/vvCa7cOzHY
81 Mac apps are on sale for Black Friday 🖤
Soundbar and Levers: 76% off
Vitals: 73% off
MightyBench: 62% off
ProtectMyMac: 60% off
76 more are waiting in one place 👇
https://t.co/vAGXWCVmWo
🚨 Jensen Huang has resigned from NVIDIA. Effective immediately.
After 33 years.
NVIDIA confirms he’s stepping down as CEO, president AND a board member.
→ CFO Colette Kress becomes interim CEO while retaining her CFO role
→ Harvey C. Jones becomes NVIDIA’s first-ever board chair
→ A permanent CEO search will consider internal + external candidates
The board says operations, customer commitments and the product roadmap are expected to continue without interruption.
But NVIDIA without Jensen leading it?
That is a HUGE change for the company at the center of the AI infrastructure race.
Earnings are still scheduled for November 17.
That call just became a lot more interesting.
$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.
Anthropic’s IPO site tells a much bigger story than the latest Claude release.
The story starts with a bet on scaling.
In 2020, four of its future founders helped write the scaling-laws paper: more compute, data and model capacity could predictably produce more capable AI.
Their company is now reporting $17 BILLION in revenue for a single quarter.
Q3 2026, specifically.
More than 12× year-over-year growth.
And over 11,900 customers each generating at least $100K in trailing 12-month revenue.
That enterprise adoption is the part that stands out to me.
Lilly has given more than 50,000 employees access to Claude Enterprise.
Legora’s Claude Code deployment grew from around 50 seats to 700.
These are examples Anthropic highlights of Claude spreading through organizations after the initial deployment.
Then there’s what’s happening inside Anthropic.
The company says Claude wrote over 90% of contributions to its codebase in September.
Each generation is helping build the next.
That’s a pretty wild development loop:
Better models help researchers and engineers move faster.
Their work produces the next generation of models.
Those models become useful for more work.
Anthropic also reports more than 4× improvement in compute efficiency since Q1 2025, measured as revenue relative to compute + infrastructure expenses.
Its infrastructure spans AWS Trainium, Google TPUs, NVIDIA GPUs and AMD GPUs.
But the financial picture needs the full context.
Q3 operating margin was -3%.
Adjusted operating margin was +15%.
Net loss was $23.3B, including a $21.9B non-cash charge tied to changes in the value of convertible financing instruments.
Those are different measures. The adjusted margin doesn’t make the reported loss disappear.
The founders’ letter is equally interesting.
All seven co-founders still work at Anthropic.
They say they’ve pledged 80% of their personal Anthropic equity to charitable causes.
The company is also setting aside 5% of its equity for a public-good fund before the IPO.
Its Long-Term Benefit Trust elects four of seven directors, with specific safeguards around one seat.
And the founders explicitly say they may make decisions that sacrifice commercial upside for safety or their public-benefit mission.
That includes delaying releases and keeping Claude free of advertising.
The proposed Nasdaq ticker is ANT.
The site currently targets pricing on November 18 and trading on November 19, subject to regulatory clearance.
My read: Anthropic is trying to connect frontier research, enterprise adoption and AI-assisted development into a business that keeps improving its own foundations.
The growth is enormous.
So is the challenge of sustaining it while keeping those commitments.
That’s the Anthropic story I’m watching.
Our initial public offering: https://t.co/Ah4BkI2loD
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: https://t.co/Z44axxzf9b) 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 https://t.co/94ojsWZTKi. 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.
🚨 Qwen4 is coming — and Qwen4-Max is already in preview.
Alibaba says the focus is stronger coding, deeper knowledge + better agentic capabilities.
That includes complex coding, front-end builds, knowledge work and professional software workflows.
Where to get access:
→ Preview on Alibaba’s Token Plan
→ Early access through QwenWork, Qoder, Qwen APP + QwenStudio
The part I’m most curious about is coding.
Can it handle a difficult repo, make the right changes and finish the job without constant correction?
Fable 5.2.
Gemini 4 Carbon.
Now Qwen4-Max.
Qwen4 is coming.
Deeper knowledge. Richer world understanding. Stronger coding and agentic capabilities.
Built for real-world tasks — from complex coding and knowledge work to front-end builds and professional software workflows.
Qwen4-Max is now in preview on Alibaba’s Token Plan. You can also get early access to the new model on QwenWork, Qoder, Qwen APP, and QwenStudio.
Put it to the test on a real problem. Show us what you build. 👀
Token Plan: https://t.co/LOB2Jdt3Sk
Anthropic just gave every Pro, Max, Ultra + Team user a FULL reset 😭
All models.
All limits.
They say they fixed issues that burned more usage than intended.
The earlier Fable resets also weren’t much help if you’d already exhausted your weekly limits across all models.
So this one resets the whole lot.
Claim it when you need it, before December 10 at 2 PM PT.
Settings → Usage
Or → https://t.co/UlvSTdIG70
Fix the issue. Give the usage back.
A massive W from @ClaudeDevs.
Google just announced Gemini for healthcare.
An agent system designed to work across patient records, insurance policies and administrative tools.
The workflows are VERY practical:
→ Summarize patient history across disconnected records
→ Prepare charts before appointments
→ Assemble prior authorization packets
→ Flag missing documentation before submission
→ Draft evidence-backed appeals
→ Explain coverage + estimated out-of-pocket costs
The interesting part is the integration.
Epic + Oracle Health connectors provide read-only access, preserving existing permissions without writing back to patient charts.
Teams can build custom healthcare skills around their own policies and procedures.
Google says the platform supports HIPAA compliance, with citations, audit logs and human oversight.
Patient data, prompts and outputs aren’t used to train Google’s foundation models.
Available in U.S. preview.
Getting the right records, evidence and policy details in front of a clinician or reviewer is a huge part of the work.
That’s where this could make a serious difference.
Today, we’re announcing Gemini for healthcare.
This newest addition to our suite of purpose-built industry solutions extends the Gemini agent to health systems and health plans.
Learn more → https://t.co/FxnlgR7uJv
🚨 Gemini 4 Carbon is now available.
Google finally shipped it.
Access:
→ Paid Gemini API in AI Studio
→ Gemini for business
→ Rolling out to Google AI Ultra users in Antigravity, the Gemini app + AI Mode in Search
Introductory API pricing:
• $2 / 1M input tokens
• $10 / 1M output tokens
That expires November 30. Afterward, pricing doubles to $4 / $20.
Google says Carbon ranks #1 on CursorBench 4.0, WANDR and AutomationBench, at less than half the cost per task of the next-best external model.
The interesting part for builders is /swarm in Antigravity.
One prompt becomes a coordinated agent team, with isolated workspaces + verifiers checking changes.
Carbon handles architecture, coordination and auditing while lighter models handle other tasks.
Google also says Carbon uses MORE tokens than Argon for deeper reasoning and extended verification.
So that introductory token price won’t tell you the whole cost story.
Fable 5.2 just landed.
Now Carbon is available with aggressive launch pricing + a multi-agent workflow.
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.
My prediction: Anthropic has a whole new Claude family coming sooner than people think.
Here’s how I think this plays out:
→ The IPO gets pushed back by a few weeks
→ The model teams use that window for more training, evaluations and final polish
→ Anthropic drops a new Claude family just before or right after the IPO
The part I keep thinking about is the timing of Fable 5.2.
My theory is that this release puts pressure on OpenAI to finally show its next Astra and Sol.
Anthropic has moved the bar. OpenAI now has to decide whether its next releases clear it.
I wouldn’t be surprised if the new Astra and Sol beat Opus 5.5 and Fable 5.2.
But I also wouldn’t assume Anthropic’s next response is months away.
What if Fable 5.2 is already buying time for the next family?
OpenAI ships.
Takes the lead.
Anthropic responds with upgrades across the Claude lineup.
A stronger flagship, a better everyday coding model, and a faster, cheaper model for agent workloads.
That would make the IPO window VERY interesting.
A fresh model family taking the lead would give Anthropic a much stronger story than defending a release everyone has already had time to catch.
This is my prediction, not a leaked roadmap.
But my bet is that Fable 5.2 is only the opening move.
The next fight for #1 could be much shorter than Opus 5.5’s 48-day run 👀
Gemini 4 Carbon is looking imminent 👀
Google’s team is hyping it hard, and the benchmark numbers suggest it could beat Opus 5.5.
But I’m more curious about the coding experience.
How well does it navigate a repo, make changes and get things working without constant hand-holding?
If Carbon can beat Sonnet 5.5 in my daily coding workflow, that’s a HUGE win for Gemini.
That would give me a reason to switch.
Do you think Carbon will actually mog Sonnet?
Fable 5.2 might have just triggered a chain reaction across the AI industry.
Everyone has a new target to beat.
The speculative response lineup is getting ridiculous:
→ OpenAI: GPT-6.2 Astra or GPT-6 Bel
→ Google: Gemini 4 Krypton
→ SpaceX: Grok 4.8
And Anthropic’s next move could be Opus 5.6 or Fable 5.5.
Those names and timelines aren’t all confirmed, so don’t treat this as a release calendar.
But imagine spending months preparing your next model, then watching Fable 5.2 move the bar right before you ship 😭
Do you release anyway?
Or go back and cook longer?
If these releases start landing together, the next few weeks are going to be INSANE.
Grok 4.8 is finally moving 👀
Elon says they scrapped JAX/SGLang and rewrote the entire RL + inference stack from scratch in C/C++.
That’s part of why it took longer.
Current status:
→ Pre-training is done
→ SFT data needed substantial cleanup
→ 4.8 is now in reinforcement learning
→ Evals are starting to climb
→ A soft rollout to Bot users is planned for later this week
The interesting part is how he’s setting expectations.
He says 4.8 will bear no resemblance to 4.7 — “thankfully” 😭
But also admits it’s “not yet incredible.”
His bet is that getting the foundations right will let it improve from here.
After 4.7, Grok badly needs this to land.
Rewriting the stack is one thing.
Now show us what 4.8 can actually do.
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.
Fable 5.2 might have just opened the release floodgates.
Rumours I’m hearing:
→ Gemini 4 Carbon could drop today
→ Qwen 4 was reportedly expected last week, so this week could be next
And I wouldn’t be surprised if OpenAI finally ships GPT-6.2 Astra or Sol this week.
Anthropic has gone first.
Everyone can see the scores they need to beat now 😭
The interesting part is whether these releases were already lined up or Fable 5.2 just made everyone reconsider what they’re about to ship.
Nothing confirmed on those dates yet.
But if even half of this lands, we’re getting an avalanche of models.
Lock tf in 👀
Anthropic dropping Fable 5.2 has finally started the release frenzy off
Gemini 4 Carbon is likely dropping today, I wouldn't be surprised if OpenAI finally release GPT-6.2 Astra or Sol this week now Anthropic have gone first and they can assess if they're screwed or not
Qwen 4 was meant to drop last week so probably got delayed to this week if I had to guess, anyway lock tf in there will be an avalanche of models soon
🚨 Anthropic just launched its IPO roadshow.
Proposed Nasdaq ticker: ANT.
The numbers:
→ 825 million shares
→ Expected price: $70–$78 per share
→ Base offering size: $57.75B–$64.35B
→ Another 123.75 million shares available through the underwriters’ option
That’s the offering size, not Anthropic’s valuation.
Of the base offering, ~801.35 million shares are newly issued. The rest come from existing stockholders.
The IPO is still proposed. Its SEC registration hasn’t become effective yet, and the final price isn’t set.
First Fable 5.2.
Now an IPO roadshow targeting roughly $58B–$64B in share sales.
Anthropic is having an absolutely insane week.
Anthropic has launched the roadshow for its proposed initial public offering: https://t.co/FhVRVuWN9X
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.
Claude Opus 5.5 stayed on top for 48 days.
Today, Fable 5.2 took its spot.
Anthropic held the lead long enough to beat itself 😭
Everyone else is still chasing.
Gemini 4 Carbon is looking dead on arrival 😭
From Anthropic��s benchmark table:
FrontierSWE v2:
Fable 5.2 → 69.9%
Carbon → 61.2%
Computer use:
Fable 5.2 → 89.3%
Carbon → 81.5%
Agentic scientific research:
Fable 5.2 → 74.6%
Carbon → 64.8%
Carbon does win on business workflows.
But those coding + computer-use gaps are rough.
And if GPT-6.2 Sol only lands around Opus 5.5?
Anthropic has already moved the target again.
Matching Opus would put it in a race for second place on a lot of this table.
It’s joever 💀
Muse Ember > Grok 4.7 😭
Meta’s smallest, cheapest model just mogged Grok on the Artificial Analysis Intelligence Index.
→ Ember (Max): 36
→ Grok 4.7 (XHigh): 35
Cost per task:
→ Ember: $0.14
→ Grok: $7.77
A higher score at roughly 55× lower cost 💀
@elonmusk you gotta cook something.
At this rate, the next Chinese model release is going to make this even funnier.
@grok where is Grok 4.8?
Fable 5.2 coordinated over 1,000 subagents for EIGHT DAYS to build an operating system 🤯
Starting from an empty folder.
Anthropic says the agents worked around the clock writing, testing and reviewing code.
The result is a prototype OS that actually boots on a laptop, with:
→ Its own kernel
→ Disk, keyboard + network drivers
→ Permanent file storage
→ Windows + a login screen
→ An installer
→ Built-in Claude functionality
An empty folder to a bootable operating system.
With Claude inside it 😭
The part that stands out to me is coordinating that much work across eight days and getting the pieces to work together.
Anthropic, what the actual hell.
OpenAI just revealed Control Room.
A central system that can cut off model output across its research infrastructure within 35 seconds of a stop instruction.
Training runs, evaluations, agentic workloads, data processing.
The interesting part is WHERE they enforce it.
At the inference layer.
Every inference request needs authorization from Control Room. Once a job is stopped, further requests are denied.
The research process can stay alive, but it loses access to model output.
And if Control Room itself goes down?
Inference requests get denied too.
That includes failures or tampering. Generation stays blocked until the service recovers.
This protection is already enabled across all frontier model research workloads, with the remaining research workloads following over the next few days.
The 35-second delay comes from authorization caching. OpenAI says it’s working to bring that down to zero.
More capable agents need reliable ways to stop them.
The engineering behind that is becoming seriously interesting.
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.
https://t.co/tc0lBqFdS5