科技热门推文——2026年7月15日

今日科技动态：人工智能扩张进一步提速——一款拥有270亿参数的多模态模型已登陆智能手机，OpenAI模型进驻亚马逊Bedrock平台，多款办公工具的用户数也增至数百万。初创企业领域再获关注，DeepSeek据报正筹划首次公开募股，同时新的开发者激励措施陆续推出。与此同时，人工智能基础设施需求推动中国芯片出口增长，但IBM软件及大型机销售表现不及预期，引发其股价遭遇历史性抛售。随着通用人工智能日益临近，围绕人工智能引发的裁员、无处不在的智能设备、自主化劳动力，以及德米斯·哈萨比斯呼吁加强治理等议题的争论持续升温。

1. danshipper (Group Score: 277.6 | Individual: 41.7)

Cluster: 8 tweets | Engagement: 317 (Avg: 176) | Type: Tech

BREAKING:

Introducing All Access from @every, our new membership tier for the best builders in AI

All Access subs get the Builder Pack which includes $7,000 in credits and free usage to the models + tool stack we use @every. All Access subscribers get:

$1,000 in Codex / @ChatGPTapp for Work credits
12 months free of @Cursor_AI Pro+
$4,000 in @PostHog credits including self-driving to automatically fix bugs and identify issues in your production app
1 year free of @Framer
6 months free of @NotionHQ

And much more! (Did I mention $1,000 in Codex credits? It's time to build!)

Get all access: https://t.co/3HlZawcc9N

Why All Access and the Builder Pack This is the best time in history to build something.

For a long time, it’s been possible to one-shot impressive demos, but they’d fall flat the minute they hit production. But the release of GPT-5.6-Sol and Fable 5 heralds a new era: Everyone can build, launch, and maintain the software that they’ve always dreamed of. Everyone is a builder now.

There’s just one catch: Building with AI is very expensive. (Ask me how I know.) (Alright, I’ll tell you. I accidentally used 2 billion tokens overnight this week on a big GPT-5.6-Sol run. Worth it.)

This is unique in the history of technology. For most of the personal computing era, a billionaire and a solo builder could buy essentially the same top-of-the-line Mac. AI changes that: The more tokens you can afford, the more you can make.

And we want to make that accessible to more people. That’s why the main feature of our new All Access plan is the Builder Pack: more than $7,000 in credits and discounts on the full stack we use to run Every, from idea to production—Codex, Claude, PostHog, Render, Gemini, FLORA, and more.

Early-bird membership is only $500/year for the next 24 hours—and the Codex credits alone are worth$ 1,000.

(I could’ve used it for my overnight run this week.)

Now we’re handing it to you.

Get all access: https://t.co/3HlZawcc9N

Meet the Builder Pack It's got more than $7,000 in offers from 10 of the AI products we use to write, design, build, and run @Every:

BUILD

$1,000 in Codex credits plus one month of ChatGPT for business
Twelve months free of Cursor Pro+
One month free of @Claudeai Max
Three months free of @Google AI Pro

DESIGN

One year free of @Framer Pro
One month free of @floraai Max

HOST

$300 in @render credits

IMPROVE

$4,000 in PostHog credits
Six months free of @Notion Business
Six months free of @AgentMail

We rely on these every day, and we tried to put together a package that helps you comprehensively for each part of the process of building and running software in AI.

What comes with All Access

Everything in an existing paid Every membership: our daily writing, guides, camps, and software like @usemonologue, @CoraComputer, @SparkleApp, and @TrySpiral
The Builder Pack, with more than $7,000 in partner offers
Unlimited email accounts use of Cora and unlimited Spiral usage
Members-only programming with me and the Every team and me

Get All Access: https://t.co/3HlZawcc9N

See 7 related tweets

@danshipper: builders love @every's builder pack! https://t.co/w6Ne8xZkGq\n\nQT @danshipper: BREAKING:

Introduci...

@bran_don_gell: Technical or non-technical. Parent or 12-year-old. Professional athlete or intramural star. Struggli...
@ajambrosino: seems big\n\nQT @danshipper: BREAKING:

Introducing All Access from @every, our new membership tier ...

@danshipper: we have now issued over 1 million codex credits to @every subscribers!!!\n\nQT @danshipper: BREAKING...
@tedescau: "If you like making stuff, now is the best time to be alive, ever."\n\nQT @danshipper: BREAKING:

In...

2. ClementDelangue (Group Score: 264.7 | Individual: 45.4)

Cluster: 11 tweets | Engagement: 610 (Avg: 91) | Type: Tech

RT @PrismML: Today, we’re announcing Bonsai 27B: the first 27B-class model to run on a phone.

Bonsai 27B is the new multimodal flagship of the Bonsai family. Based on Qwen3.6 27B, it brings a new capability tier to local AI: multi-step reasoning, structured tool use, long-context workflows, and coherent agentic loops.

Until now, models in this class have been impractical to deploy locally. A 27B model occupies roughly 54 GB in 16-bit precision, and even a strong 4-bit build is around 18GB - too large for a phone and for most laptops. Bonsai 27B changes that.

It comes in two variants: • Ternary Bonsai 27B: 5.9 GB, 1.71 effective bits per weight, optimized for laptop-class quality. • 1-bit Bonsai 27B: 3.9 GB, 1.125 effective bits per weight, optimized for phone-class footprint.

Everything is open-sourced today under the Apache 2.0 license.

See 10 related tweets

@TeksEdge: Oh 💩! Local AI just leveled. Bonsai 27B is the first real 27B-class model that actually runs in 4–6...
@HessianFree: We did a thing.

We just launched Bonsai 27b Binary and Ternary models based on Qwen 3.6. Apache 2.0...

@scaling01: A 27B model using a mere 3.9GBs https://t.co/bonCTy1tEt\n\nQT @PrismML: Today, we’re announcing Bons...
@VaibhavSisinty: This is insanee 🤯

Finally. A 27B parameter AI model running on a phone.

PrismML just shipped Bonsa...

@HessianFree: RT @TeksEdge: Oh 💩! Local AI just leveled. Bonsai 27B is the first real 27B-class model that actual...

3. unusual_whales (Group Score: 245.4 | Individual: 28.7)

Cluster: 16 tweets | Engagement: 17084 (Avg: 2692) | Type: Tech

Meta, $META, reportedly used AI to target workers with medical conditions for layoffs, per Reuters

See 15 related tweets

@HedgieMarkets: 🦔File this one under "we built the AI, we swear we didn't use it." 26 former Meta employees are suin...
@DeItaone: META USED AI-POWERED SOFTWARE TO TARGET WORKERS WITH MEDICAL CONDITIONS FOR LAYOFFS, LAWSUIT CLAIMS...
@CNBC: Employees sue Meta, alleging discrimination in using AI to make layoffs https://t.co/xoYbsIOdcH...
@Reuters: Meta used AI to target workers with medical conditions for layoffs, former employees' lawsuit claims...
@MorePerfectUS: Meta used AI to target workers with medical conditions for layoffs, according to a new lawsuit.

Twe...

4. wallstengine (Group Score: 221.3 | Individual: 32.0)

Cluster: 9 tweets | Engagement: 427 (Avg: 148) | Type: Tech

$IBM released preliminary Q2 results BELOW internal expectations as mainframe and software sales disappointed, with large deals failing to close on expected timelines.

CEO: “This quarter we faltered.”

Revenue: $17.2B vs$ 17.86B est. 🔴, up 1% YoY Software: +5% Consulting: flat, +1% cc Infrastructure: -7% Non-GAAP EPS: $2.93, up 5% YTD free cash flow:$ 4.8B

CEO Arvind Krishna said: “In the last few weeks of June, we saw clients shift their quarterly capex spend toward servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases.”

He also said clients were distracted by “rapidly-evolving, industry-wide cybersecurity concerns,” and that IBM “did not anticipate the magnitude of the capex reprioritization.”

Full Q2 results are scheduled for July 22.

See 8 related tweets

@StockSavvyShay: $IBM is down more than 23% after revenue missed estimates driven by “weakness in the software and in...
@StockMKTNewz: $IBM JUST REPORTED PRELIMINARY Q2 EARNINGS AND ITS DOWN 15% 🔴

IBM just said for Q2 it brought in

...

@StockSavvyShay: $IBM is down more than 25% marking its largest drop in nearly 60 years. https://t.co/lWb84sGxtG\n\nQ...
@amitisinvesting: $IBM

Woah.

IBM put out preliminary Q2 results...and they are not good.

Revenue was $17.2B vs$ 17....

@StockMKTNewz: If you look at it from this perspective, it looks like things couldn’t be going better 🫡 https://t.c...

5. jxnlco (Group Score: 207.4 | Individual: 32.7)

Cluster: 9 tweets | Engagement: 1318 (Avg: 262) | Type: Tech

RT @thsottiaux: Hello. We have reached 8M active users across Codex and ChatGPT Work.

We are once again resetting the usage limits for all. And we continue to not have the 5h rate limit as well, allowing everyone to explore the boundaries of GPT-5.6 Sol and discover how ambitious you can be.

See you tomorrow for more updates on our growth!

See 8 related tweets

@thsottiaux: Hello. We have reached 8M active users across Codex and ChatGPT Work.

We are once again resetting t...

@ajambrosino: reposting my friend’s tweet for some visibility\n\nQT @thsottiaux: Hello. We have reached 8M active ...
@sama: hello!\n\nQT @thsottiaux: Hello. We have reached 8M active users across Codex and ChatGPT Work.

We ...

@shiri_shh: Holy f*ck.

OpenAI just added 1 MILLION active users in the last 24 hours.

Sam and Tibo are comple...

@derrickcchoi: Awesome seeing this momentum and users loving the product. 🚀\n\nQT @thsottiaux: Hello. We have reach...

6. KobeissiLetter (Group Score: 201.1 | Individual: 24.3)

Cluster: 11 tweets | Engagement: 3243 (Avg: 3436) | Type: Tech

BREAKING: DeepSeek is preparing for an IPO and may file as soon as this year, per Bloomberg.

Details include:

The company is reportedly in talks with accounting and banking advisors and the IPO is expected to be listed in China
DeepSeek has begun talks for a fresh funding round valuing the company at $71 billion
Amid rumors of an IPO, DeepSeek is now looking to expand into the agentic AI market

The AI arms race is accelerating.

See 10 related tweets

@wallstengine: DeepSeek is already weighing a new funding round with a 37% valuation jump, just one month after rai...
@minchoi: DeepSeek is preparing to file an IPO this year https://t.co/YKWnkM3NTJ\n\nQT @WatcherGuru: JUST IN: ...
@negligible_cap: *DEEPSEEK IS SAID TO PREPARE FOR IPO FILING AS SOON AS THIS YEAR

DeepSeek IPOing before OpenAI? htt...

@coinbureau: ⚡️UPDATE: DeepSeek is preparing for a mainland China IPO, targeting a filing this year and a potenti...
@business: Chinese AI pioneer DeepSeek has begun preparations for an initial public offering and may file as so...

7. negligible_cap (Group Score: 190.6 | Individual: 33.3)

Cluster: 9 tweets | Engagement: 34 (Avg: 68) | Type: Tech

Call me crazy but this OpenAI product sounds pretty stupid so long as you own a smartphone. Except for the fact it can constantly watch and listen to you in your home.

“It includes a camera and other sensors that help it understand a user’s surroundings and context, as well as advanced AI models beyond those available on conventional smart speakers. “

“Another central difference is that the device includes a rechargeable battery, allowing it to be carried from room to room throughout the day. A user could bring it into the laundry room while doing chores, move it into the kitchen for cooking assistance, and later place it in a living room or bedroom to have it play music.”\n\nQT @negligible_cap: *OPENAI PLANS TO RELEASE HOME SPEAKER AS FIRST HARDWARE PRODUCT

Now that's a curveball

See 8 related tweets

@unusual_whales: BREAKING: OpenAI is planning to release a screenless ‌speaker as its first hardware product, per B...
@negligible_cap: *OPENAI PLANS TO RELEASE HOME SPEAKER AS FIRST HARDWARE PRODUCT

Now that's a curveball...

@FirstSquawk: OPENAI PLANS TO RELEASE HOME SPEAKER AS FIRST HARDWARE PRODUCT, DESIGNED AS AI COMPANION...
@exec_sum: BREAKING: OpenAI’s first device will be a movable, screenless smart speaker built as an AI companion...
@financialjuice: OpenAI plans to release a home speaker as its first hardware product, designed as an AI companion....

8. gdb (Group Score: 162.8 | Individual: 35.4)

Cluster: 7 tweets | Engagement: 628 (Avg: 1041) | Type: Tech

GPT-5.6 is now generally available on Bedrock:\n\nQT @AWSNewsroom: GPT-5.6 Sol, Terra, and Luna from @OpenAI are now generally available on Amazon Bedrock. Three tiers of intelligence, from flagship reasoning to fast inference, running on Bedrock's next-generation inference engine built for high-performance, security, scale, and reliability. https://t.co/Hn2z5M2Jp2

See 6 related tweets

@TheRealAdamG: GPT-5.6 🤝 AWS Bedrock\n\nQT @AWSNewsroom: GPT-5.6 Sol, Terra, and Luna from @OpenAI are now generall...
@OpenAIDevs: RT @AWSNewsroom: GPT-5.6 Sol, Terra, and Luna from @OpenAI are now generally available on Amazon Bed...
@VKazulkin: "OpenAI GPT-5.6 Sol, Terra, and Luna now generally available on Amazon Bedrock". This brings the sma...
@jxnlco: GPT-5.6 Sol, Terra, and Luna are now available on Amazon Bedrock.

Use the Responses API in AWS, wit...

@VKazulkin: RT @SwamiSivasubram: Today, we’re bringing customers the most powerful models from @OpenAI —GPT-5.6 ...

9. DataChaz (Group Score: 152.6 | Individual: 69.7)

Cluster: 4 tweets | Engagement: 5113 (Avg: 359) | Type: Tech

✅ Claude Code ✅ Claude Cowork ✅ Claude Design ✅ Claude Finance ✅ Claude Science ✅ Claude Teacher ⬜ Claude HR ⬜ Claude Analytics ⬜ Claude Marketing ⬜ Claude Sales ⬜ Claude Legal ⬜ Claude R&D ⬜ Claude Procurement ⬜ Claude Accounting ⬜ Claude Engineering https://t.co/q522bSCgxk\n\nQT @claudeai: We're introducing Claude for Teachers: free access to premium Claude capabilities for verified K-12 educators in the US, with a library of teaching skills and a direct connection to evidence-based curricula, mapped to academic standards in all 50 states.

https://t.co/5hZZijVPCV https://t.co/5ofG8YLEON

See 3 related tweets

@cgtwts: Teachers realizing they just got Claude Pro for free: https://t.co/N5dJnRr7p2\n\nQT @claudeai: We're...
@testingcatalog: Anthropic announced Claude for Teachers, a new free solution for verified K-12 teachers in the US. ...
@DataChaz: RT @DataChaz: ✅ Claude Code ✅ Claude Cowork ✅ Claude Design ✅ Claude Finance ✅ Claude Science ✅ Clau...

10. sairahul1 (Group Score: 145.4 | Individual: 41.5)

Cluster: 6 tweets | Engagement: 228 (Avg: 104) | Type: Tech

Jacob Bank, former Google product lead:

"I built up this team of 40 AI marketing agents to work with me. I'm the only marketing person."

In a 15-minute talk, he shows what one person with the right setup now runs alone.

Forty agents. One human. His AI bill is $500 a month, against the$ 50,000 a human team would cost.

That's the math quietly minting the first solo fortunes of the AI era.

Watch the talk, then read the article to build your first AI team.\n\nQT @sairahul1: https://t.co/D9LpnywGnl

See 5 related tweets

@Av1dlive: Google pays AI engineers $450,000 a year to build AI models

That same Google Engineer just dropp...

@eng_khairallah1: RT @sairahul1: Jacob Bank, former Google product lead:

"I built up this team of 40 AI marketing ag...

@Av1dlive: RT @Av1dlive: Google pays AI engineers $450,000 a year to build AI models

That same Google Engine...

@eng_khairallah1: RT @sairahul1: Sam Altman:

"We're going to see 10-person billion-dollar companies pretty soon."

Mo...

@heynavtoor: RT @heynavtoor: "Mindblowing all-in-one AI which does everything in a single solution."

That is Ant...

11. Reuters (Group Score: 137.3 | Individual: 24.8)

Cluster: 7 tweets | Engagement: 74 (Avg: 130) | Type: Tech

China's exports surged in June, buoyed by demand for chips and data center computing power to fuel the global AI boom, giving policymakers grappling with how to boost domestic demand in the world's second-largest economy a much-needed buffer https://t.co/cRYjwoykiS

See 6 related tweets

@Reuters: China's exports surged in June, buoyed by orders for chips to fuel the global AI boom and automobile...
@ReutersBiz: China's exports surged in June, buoyed by orders for chips to fuel the global AI boom and automobile...
@ReutersBiz: China's exports surged in June, buoyed by demand for chips and data center computing power to fuel t...
@business: China’s exports and imports expanded faster than forecast in June, as surging chip prices and global...
@CNBC: China exports jump at fastest pace since 2021 as AI boom, tariff rush lift trade https://t.co/IRp07l...

12. cryptopunk7213 (Group Score: 130.7 | Individual: 29.1)

Cluster: 6 tweets | Engagement: 28 (Avg: 51) | Type: Tech

i want to see more of this! AI should absolutely replace all the boring administrative tasks that add zero value to a humans experience on earth.

ironically this results in much better education and research https://t.co/8JglGidJXj\n\nQT @claudeai: We're introducing Claude for Teachers: free access to premium Claude capabilities for verified K-12 educators in the US, with a library of teaching skills and a direct connection to evidence-based curricula, mapped to academic standards in all 50 states.

https://t.co/5hZZijVPCV https://t.co/5ofG8YLEON

See 5 related tweets

@VadimStrizheus: teachers in school

“don’t use AI for anything, it’s cheating”

Anthropic just now

“Launching free ...

@EHuanglu: teachers realising AI can teach any subject 24/7 https://t.co/fTzaY8yK0g\n\nQT @claudeai: We're intr...
@adocomplete: RT @drew_bent: Today we're launching Claude for Teachers -- premium @claudeai and Cowork, free for e...
@Polymarket: JUST IN: Anthropic unveils “Claude for Teachers,” offering free premium AI tools for U.S. K-12 educa...
@RoundtableSpace: ANTHROPIC LAUNCHES “Claude for Teachers”

Your kids will soon be learning a curriculum created by AI...

13. cryptopunk7213 (Group Score: 124.7 | Individual: 35.9)

Cluster: 5 tweets | Engagement: 39 (Avg: 51) | Type: Tech

demis calling for a new AI organization responsible for governing new models going forward, some highlights:

the org would create a protocol for assessing new models 30 days before release, ai labs would be required to share model cards etc before.
model review group composed of frontier ai researchers from labs + open source.

-ai industry will fund the effort in order to attract the smartest talent + pay for the compute to test new models throroughly.

america should be the ones to create and own this as the leading country for AI growth.
if a model is deemed a national security risk then model release will be slowed down.

basically sounds like what the government is already doing but in a more formal body like FINRA or the SEC and way more transparency,.

well written and worth the read\n\nQT @demishassabis: https://t.co/PTeDiv1b6L

See 4 related tweets

@AnandButani: AGI is only a few years away.

The current intense commercial/geopolitical race is outpacing our un...

@Techmeme: Demis Hassabis proposes a US-based Standards Body for "Frontier-class" AI, modeled after the FINRA; ...
@Hesamation: Demis Hassabis (Google DeepMind) asks the government to step in on frontier AI, or even order a slow...
@baseten: We're seeing it too: companies of all sizes are adopting more open-source AI, including alongside cl...

14. KobeissiLetter (Group Score: 115.0 | Individual: 31.9)

Cluster: 5 tweets | Engagement: 3666 (Avg: 3436) | Type: Tech

It's official:

IBM stock, $IBM, is now on track for its biggest daily decline since 1968, down -26%.

That's -$70 billion in market cap erased today.

The lesson? Don't get comfortable in the AI economy, keep evolving. https://t.co/K5CM5uIZu0

See 4 related tweets

@KobeissiLetter: BREAKING: IBM stock, $IBM, collapses nearly -20% after posting weaker than expected earnings due to ...
@KobeissiLetter: The biggest lesson of the AI era: never get comfortable.

On June 2nd, IBM, $IBM, was up +13% for th...

@shanaka86: AI triggered a run on hardware. IBM says clients bought tomorrow's servers and memory today, before ...
@FastCompany: ‘We faltered’: IBM stock collapses after a grave warning about AI https://t.co/TulqJPMTEv...

15. heyshrutimishra (Group Score: 110.2 | Individual: 46.8)

Cluster: 3 tweets | Engagement: 309 (Avg: 94) | Type: Tech

Demis Hassabis the CEO of Google DeepMind just published an essay calling AGI a "few short years away" and comparing it not to the internet or mobile, but to the discovery of fire.

Then he wrote this:

"Nobody in the world knows for sure what is going to happen from here, and even the experts disagree."

Read it again.

His estimate on impact: 10x the Industrial Revolution at 10x the speed.

And then he proposes a fix...

A US Standards Body modeled on FINRA, the financial regulator that governs Wall Street. Funded by industry. Labs share models 30 days before release. Once the protocol works, it becomes mandatory: a model must pass to ship in the US market. Agentic systems tested specifically for deception and guardrail bypass. Benchmarks updated quarterly.

The framework applies to any frontier model regardless of country of origin. Chinese or American. Open or closed.

That line is doing more work than it looks like....

DeepSeek claimed to train a GPT-4 level model for $6 million and open-sourced it to the world.
Alibaba's Qwen now commands over 50% of global open-source model downloads.
ByteDance's Doubao crossed 100 million daily users.

The US is not building this alone.

A Standards Body with no reach into Chinese models has a very large hole in it. US market access as a compliance condition is the only lever that actually closes it.

Here is the tension the essay does not resolve...

Google DeepMind is one of the labs racing hardest toward the thing Hassabis says nobody understands. Alphabet spent $91 billion on capital expenditure in 2025. He is simultaneously pressing the accelerator and asking someone else to build the brakes.

He has been right before. Neural networks when academia said they were dead. Games as the ladder to science. AlphaFold solving what biology could not solve in 50 years. Nobel Prize, 2024. His track record on long contrarian bets is not average.

Which is why the admission buried in this essay matters more than the proposal. The person most likely to know what comes next just wrote publicly that nobody knows what comes next.

A 30-day review window. For something he compares to the discovery of fire.

That is either the most important policy proposal in AI this year or evidence that even the people building this have no idea how to slow it down.

I think it is both.\n\nQT @demishassabis: https://t.co/PTeDiv1b6L

See 2 related tweets

@VaibhavSisinty: Demis Hassabis just published a framework for how AGI should enter the world. And it reads like a wa...
@heyshrutimishra: RT @heyshrutimishra: Demis Hassabis the CEO of Google DeepMind just published an essay calling AGI a...

16. samdblond (Group Score: 107.0 | Individual: 37.7)

Cluster: 3 tweets | Engagement: 79 (Avg: 29) | Type: Tech

How to accelerate revenue growth at your startup.

The future of sales and GTM in the AI era.

And more.

With the 🐐 @jaltma\n\nQT @jaltma: I sat down with @samdblond, founder of Monaco, to talk about to talk about the future of sales and GTM in the AI era.

He's an incredible GTM leader and founder and I get new ideas every time I talk to him. Hope you enjoy.

Timestamps: (0:00) Intro (1:26) Sam's sales journey (3:11) EchoSign learnings (5:28) Building Zenefits GTM (7:02) Parker's 0 to $10M thought exercise (9:51) Creating a demand-rich environment (14:20) People, brand, and revenue ops (21:19) Why Sam went to Founders Fund (28:00) Starting Monaco (29:40) Salesforce’s innovator's dilemma (31:40) Going straight for the system of record (34:30) Disrupting labor (41:10) Pricing in an AI-native environment (44:28) The Monaco GTM playbook (50:06) The$ 60,000 plane experiment (54:21) Creative campaigns vs. paid advertising (1:00:06) Monaco’s product approach (1:03:13) Advice for founders just starting to sell (1:07:58) Abundant pipeline and the right to say no

See 2 related tweets

@jaltma: Something Sam learned with Parker Conrad is the importance of creating a demand-rich GTM environment...
@jaltma: Sam makes the case for why startups should invest much more time and capital into brand, and lays ou...

17. business (Group Score: 106.6 | Individual: 36.3)

Cluster: 4 tweets | Engagement: 335 (Avg: 71) | Type: Tech

DeepSeek founder Liang Wenfeng’s net worth more than doubled after his firm’s most recent fundraising round, making the Chinese entrepreneur the world’s richest among creators of AI models.

Liang is now worth $36 billion, according to the Bloomberg Billionaires Index. Read more: https://t.co/3ekPgzHLbr

📷️: VCG/Getty Images

See 3 related tweets

@MTSlive: SITUATION EXPLAINED: DeepSeek is prepping an IPO.

• DeepSeek is in talks with new backers targeting...

@business: DeepSeek founder Liang Wenfeng’s net worth more than doubled after his firm’s most recent fundraisin...
@exec_sum: NEWS: DeepSeek founder Liang Wengfeng’s net worth increased to $36 billion after the AI company’s la...

18. higgsfield_ai (Group Score: 106.0 | Individual: 35.4)

Cluster: 4 tweets | Engagement: 327 (Avg: 179) | Type: Tech

Our Chief Motion Designer built this feature for himself first.

He developed it end to end, and it proved itself in our internal content production.

This is what a tool made by creatives for creatives looks like.\n\nQT @higgsfield: Higgsfield MCP/Plugin for After Effects just got upgraded. Powered by Fable 5.

Build your own AE plugins
Drop in your reference and it decomposes into editable vector layers you can animate
AI assistant that answers or does it for you
Localize a project seamlessly https://t.co/WL4pqMgLEM

See 3 related tweets

@aakashgupta: You can now use Claude inside After Effects.

Motion design ran on hand-set keyframes for 33 years. ...

@dr_cintas: Fable 5 just moved into After Effects.

Higgsfield’s upgraded AE plugin lets you build your own plug...

@AngryTomtweets: You can now use Claude inside Adobe After Effects via Higgsfield Supercomputer 🤯

Build animations, ...

19. BitcoinNews (Group Score: 104.1 | Individual: 23.7)

Cluster: 6 tweets | Engagement: 212 (Avg: 123) | Type: Tech

JUST IN: 🇺🇸 Fed Chairman Warsh told the House Financial Services Committee that the economy's benefits from AI are not yet fully known, but monitoring implications for employment and inflation.

"That which we're now calling AI investment will soon just be called investment." 💬 https://t.co/1Y6psnPSwv

See 5 related tweets

@BitcoinNews: JUST IN: 🇺🇸 Fed Chair Kevin Warsh said AI investments present a huge opportunity with risks, adding ...
@FoxNews: AI investment is moving so fast, Fed Chair Kevin Warsh says it may soon stop being its own category....
@TFTC21: Fed Chair Warsh says AI investment is growing so fast it will soon "just be called investment," with...
@CNBC: Warsh promises inflation will be a 'thing of the past,' cites benefits of AI investment boom https:/...
@TFTC21: Fed Chairman Kevin Warsh told Congress on Tuesday that AI is "the most significant change in our eco...

20. elonmusk (Group Score: 103.9 | Individual: 40.1)

Cluster: 3 tweets | Engagement: 15594 (Avg: 7656) | Type: Tech

Grok 4.5 reaches #1 position on Long-Horizon Terminal-Bench\n\nQT @tetsuoai: The Long-Horizon Terminal-Bench paper landed around May and concluded that the results showed headroom for improvement. The best of the 15 models they tested finished seven of the 46 tasks, and the mean across all models was about two. That ceiling is what fifth place looks like on the current board.

Grok 4.5 is now at 13, and Fable 5 is at 12. A single task costs around 9.9M tokens, 231 episodes, and 85 minutes of wall clock time. That means agents are holding a plan across all of it and finishing, and that capability nearly doubled in two months.

SpaceXAI is on top, and they marketed the 4.2x output token efficiency, which undersells it. Two dollars in, six out, per million. On a benchmark where one task burns ten million tokens, the bill is dominated by input replay, and they say Grok 4.5 solves tasks in under half the number of steps, so there is less accumulated context to resend on every call. The efficiency compounds on the input side, which is the side that costs money.

Fable 5 is one task behind. Their own launch chart has them losing DeepSWE 1.1 to Fable by 17 points, and Grok 4.20 sits on this same board at 0.080 with zero completions, so whatever happened in 4.5 is not a family trait.

My read is that the 4.5 jump came out of training alongside Cursor, which is a stream of real agentic edit trajectories nobody else has at that volume, and nothing in the counterevidence argues against it compounding into the next checkpoint.

See 2 related tweets

@XFreeze: Grok 4.5 is in a league of its own when it comes to efficiency. Nothing even comes close

It deliver...

@teslaownersSV: GROK 4.5 RANKS #1 FOR REAL-WORLD CODING QUALITY AND POLISH

In a direct head-to-head test where Grok...