热门科技推文——2026年7月13日

今日科技动态：人工智能智能体成为讨论焦点。Anthropic 和 OpenAI 暂时扩大了访问范围并提高使用限额，吴恩达则预测，具备自我改进能力的系统或将很快取代传统的提示词交互方式。工程领域的关注点集中在 Grok 于真实场景编程基准测试中的进步，以及谷歌新推出的智能体开发课程。与此同时，初创企业正积极布局训练数据、智能体间支付和人工智能原生应用分发。除人工智能外，据报道，星巴克正转向自主开发定制软件，凸显出企业对掌控核心技术基础设施的兴趣日益浓厚。

1. willccbb (Group Score: 811.6 | Individual: 45.5)

Cluster: 44 tweets | Engagement: 844 (Avg: 92) | Type: Tech

thank you tibo\n\nQT @claudeai: We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.

See 43 related tweets

@MLStreetTalk: RT @claudeai: We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Cod...
@yacineMTB: If sol wasn't out this wouldn't have happened. Thanks Sam!\n\nQT @claudeai: We're extending Claude F...
@zephyr_z9: 🤣🤣🤣🤣 Feeling the pressure from Sol\n\nQT @claudeai: We're extending Claude Fable 5 access on all pai...
@DavidKPiano: You will never guess what's going to happen on July 19\n\nQT @claudeai: We're extending Claude Fable...
@petergyang: Thank you Claude :) Just extend it permanently yeah?\n\nQT @claudeai: We're extending Claude Fable 5...

2. pvncher (Group Score: 293.1 | Individual: 61.1)

Cluster: 8 tweets | Engagement: 2488 (Avg: 173) | Type: Tech

RT @thsottiaux: Morning. The last 48 hours of Codex and ChatGPT Work have been intense! Three important updates:

Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans
Rolling out changes that will make GPT 5.6 Sol more efficient across the board and that will be reflected in less usage being used so that it can take you further. Exact impact to be quantified and shared
We hit 6M active users, and are landing a usage reset in the next hour

Go do things

See 7 related tweets

@elvissun: anthropic: we reset the usage limit

tibo: we removed the usage limit

🐐\n\nQT @thsottiaux: Morning....

@VadimStrizheus: The only difference between you and Tony Stark was the 5 hour Codex limit. https://t.co/QlB20AMnb4\n...
@banteg: no more 5 hour limit in codex\n\nQT @thsottiaux: Morning. The last 48 hours of Codex and ChatGPT Wor...
@Dimillian: 🤯\n\nQT @thsottiaux: Morning. The last 48 hours of Codex and ChatGPT Work have been intense! Three i...
@Hesamation: I will reset the hell of you. I will reset the limits so much, they’d be meaningless. https://t.co/...

3. BrianRoemmele (Group Score: 104.8 | Individual: 29.3)

Cluster: 6 tweets | Engagement: 44 (Avg: 439) | Type: Tech

IN PART 2 OF X402 PAYMENTS: HOW TO GET PAID FOR YOUR DATA BY SELLING IT TO AI AGENTS. THE NEW AGENTIC ECONOMY…

I will show one absolutely astonishing articles 100s of AI Agents are buying from me.

You will not believe it!

I will brainstorm for you on things AI Agents like.

Out soon, Become a member now.

Don’t guess, know…\n\nQT @BrianRoemmele: I HAVE CONFESSION:

I HAVE BEEN WRITING ARTICLES LIKE I POST HERE FOR AI AGENTS TO LEARN FROM—A N D—THEY PAY ME!

I get cash via USDC paid by the AI Agents to me via x402.

Read the article below how I did and how you can too! https://t.co/8FCE1Rgko4

See 5 related tweets

@BrianRoemmele: “Yesterday I set up my Wordpress site like you did and you’re right, I know it’s not much, I made 3¢...
@BrianRoemmele: I HAVE CONFESSION:

I HAVE BEEN WRITING ARTICLES LIKE I POST HERE FOR AI AGENTS TO LEARN FROM—A N D—...

@BrianRoemmele: RT @BrianRoemmele: Meet x402 AI Agent payments. It has 5 times more buyers than sellers on x402 mark...
@BrianRoemmele: RT @BrianRoemmele: I HAVE CONFESSION:

I HAVE BEEN WRITING ARTICLES LIKE I POST HERE FOR AI AGENTS T...

@BrianRoemmele: RT @BrianRoemmele: Podcast: The Brian Roemmele Interview-Accepting x402 AI Payment From Agents.

Wha...

4. ns123abc (Group Score: 97.5 | Individual: 37.1)

Cluster: 4 tweets | Engagement: 11426 (Avg: 1396) | Type: Tech

RT @elonmusk: @sama We start flying them next year. Maybe you can come see them if your parole officer approves.

After stealing an open source AI charity, you then stole all of Apple’s phone technology! Wow.

What do you plan for an encore? That’s tough to beat.

See 3 related tweets

@WOLF_Financial: Elon Musk said SpaceX $SPCX will begin sending rockets to space with the intention of starting the p...
@niccruzpatane: SpaceX AI Datacenters start flying next year. 😎 https://t.co/7bpsNN5J6w\n\nQT @elonmusk: @sama We st...
@shanaka86: Every model becomes dangerous at the exact point it forgets where its authority ends!!\n\nQT @elonmu...

5. Dagnum_PI (Group Score: 93.6 | Individual: 61.0)

Cluster: 2 tweets | Engagement: 844 (Avg: 78) | Type: Tech

RT @lukepierceops: Starbucks spends $400 million a year on software. Yesterday they announced they're moving off IBM and Microsoft to build their own custom systems in-house.

IBM dropped 3% and Salesforce dropped 4% on the news.

And honestly this is, unequivocally, the biggest signal I've seen since OpenAI and Anthropic launched their consulting arms back in Q1. The largest companies in the world are done paying for software that half fits how they work.

We saw this coming about a year ago. Moved everything we build off Airtable and low-code tools and went fully custom. Already paying off, and it's only going to compound from here.

This is the opportunity right now.

You get all of a company's data into one system. You build out a single operating system for the entire business. You cut out bad, redundant processes. Then you layer AI on top of it, under the correct processes.

That's the core of AI consulting. Helping companies actually operate better.

There are a lot of fly-by-night offerings circulating right now when it comes to Ai Services.

For example, 'second brains'.

Throwing scattered data into a second brain while the processes underneath stay broken does nothing. The companies who will absolutely destroy their competition over the next 5 years are rebuilding how they work from the ground up.

Starbucks is showing you what other companies will be doing over the next several years.

Your job is to position yourself to facilitate that process for as many companies as you can.

See 1 related tweets

@diptanu: 400 million on software doesn’t feel like a lot of money for a business like Starbucks?\n\nQT @lukep...

6. adamlyttleapps (Group Score: 93.0 | Individual: 43.8)

Cluster: 4 tweets | Engagement: 641 (Avg: 73) | Type: Tech

Anthropic has an issue:

I won’t go back to opus after using fable for a solid week

They’ve shown me what a true AI feels like. I’m sold.

If it gets removed from my plan I may switch back to OpenAI if GPT 5.6 sol is that good

And surely I’m not the only one thinking the same

See 3 related tweets

@AlexFinn: BOOM! Fable 5 staying on subs. I called this yesterday

Anthropic was going to concede the AI race t...

@RobertHaisfield: “Fable is exactly the Opus 3 reincarnation” 😍😍😍\n\nQT @teodorio: Fable is exactly the Opus 3 reincar...
@AlexFinn: RT @AlexFinn: Fable 5 is a revolutionary model. I just don't understand how it survives.

When my us...

7. AnatoliKopadze (Group Score: 85.4 | Individual: 38.9)

Cluster: 3 tweets | Engagement: 679 (Avg: 427) | Type: Tech

Google Brain founder, Andrew Ng:

"100% of my tasks are done by ai agents, self-improving loops are next.

Give it 3-6 months and prompting is gone."

31 minutes of clear explanation on building self-improving agents from scratch.

Worth more than any $500 agentic course.

Watch it, then read the full guide on loops below.\n\nQT @AnatoliKopadze: https://t.co/eq0oL4fIaR

See 2 related tweets

@AnatoliKopadze: Google just dropped a 1-hour course on agentic engineering from scratch:

00:00 - Build your first A...

@RoundtableSpace: Anthropic has just published a free course on Loop Engineering

If you use AI agents, bookmark it

h...

8. bubbleboi (Group Score: 72.9 | Individual: 26.5)

Cluster: 3 tweets | Engagement: 95 (Avg: 101) | Type: Tech

Just as valuable and probably more impactful to model capabilities than compute\n\nQT @deedydas: Every single startup selling AI Training Data (July 2026)

>50 cos sell data and RL environments to big AI labs and drive AI progress behind the scenes.

They total ~ $8.5B in rev and ~$ 100B in valuation, >75% of which are just 4 players: Scale, Surge, Mercor and Handshake. https://t.co/qO52PoiE9X

See 2 related tweets

@lukas_m_ziegler: RT @deedydas: Every single startup selling AI Training Data (July 2026)

>50 cos sell data and RL...

@prz_chojecki: https://t.co/IAD1k3t3Rq is missing but otherwise great list!\n\nQT @deedydas: Every single startup s...

9. SIGKITTEN (Group Score: 72.8 | Individual: 43.9)

Cluster: 2 tweets | Engagement: 844 (Avg: 130) | Type: Tech

RT @thsottiaux: If you aren't yet bold enough to install the Codex app, you can stay in the presence of your orange crab and point it at GPT 5.6 Sol. Takes 5 minutes. Kudos to Theo for explaining one of the ways to get this done.

Step 1: Install CLIProxyAPI Step 2: Connect Step 3: Define following alias and enjoy claudex

alias claudex='CLAUDE_CODE_SUBAGENT_MODEL=gpt-5.6-sol \
CLAUDE_CODE_ALWAYS_ENABLE_EFFORT=1 \
CLAUDE_CODE_MAX_TOOL_USE_CONCURRENCY=3 \
ENABLE_TOOL_SEARCH=false \
claude --model gpt-5.6-sol'

If this gets blocked, I owe you a reset.

See 1 related tweets

@thsottiaux: If you aren't yet bold enough to install the Codex app, you can stay in the presence of your orange ...

10. sairahul1 (Group Score: 71.7 | Individual: 37.1)

Cluster: 2 tweets | Engagement: 546 (Avg: 105) | Type: Tech

RT @sairahul1: Google just dropped a 1-hour course on agentic engineering from scratch:

00:00 – How to build your first AI agent 08:24 – Build agent memory (short, persistent, long) 28:34 – Agentic loops, long-running AI agents 40:04 – How to build MCP (MCP vs API) 1:00:22 – Multi-agentic systems

This 1-hour watch will replace 10 paid agentic courses on the internet.

Bookmark this. Watch this weekend.

See 1 related tweets

@sairahul1: RT @sairahul1: Andrew Ng just dropped a 3-hour course on how to become an AI engineer in 2026:

00:0...

11. Searxly (Group Score: 71.0 | Individual: 36.0)

Cluster: 2 tweets | Engagement: 16 (Avg: 12) | Type: Tech

Introducing Searxly Support Console & Governance.

We are pleased to announce the launch of the Searxly Support Console and Searxly Governance. Both services now operate under a single unified platform at https://t.co/WUxLVn9ZGu, delivering a seamless experience for support requests and community participation in product direction.

The Support Console provides the primary channel for users to create, submit, and track support tickets. An account is required to access this functionality. The account creation process has been intentionally designed for maximum simplicity and security. Users simply provide their email address. A one time use code valid for ten minutes is then sent to that address. Entering the code activates the connection instantly. No password is ever required or stored.

The Governance system runs alongside the Support Console on the same platform. Once connected through the shared account, users gain the ability to submit suggestions that appear in the dedicated Governance tab at https://t.co/hsJl5JoeqY.

These suggestions are visible within the community and open for discussion as well as voting. Standard connected users receive an allocation of five votes together with the capacity to submit one suggestion.

Users who connect a wallet holding at least 5 $of$ SEARXLY unlock enhanced privileges. Their tickets receive priority handling within the Support Console. They also benefit from unlimited suggestions and an unlimited number of votes in Governance.

This combined system ensures accessible support for all users while providing meaningful avenues for community input and rewarding active token holders with greater influence and expedited assistance.

See 1 related tweets

@myrhex: If you want to decide the future of Searxly, join us\n\nQT @Searxly: Introducing Searxly Support Con...

12. nathanbenaich (Group Score: 69.1 | Individual: 35.1)

Cluster: 2 tweets | Engagement: 19 (Avg: 11) | Type: Tech

no time to waste\n\nQT @StartupCltn: 📣 Today with @Founders_Forum we've published an open letter to Andy Burnham: Rebuild Britain.

It's signed by 100+ founders & investors who are clear that the next few years will make or break Britain as a tech power. We must:

Back founders on tax, investment and more Unlock scaleup capital Bulldoze "computer says no" regulation Deliver procurement £ for British tech Ensure tech talent in Government

We're ready to rebuild Britain. And want to work with anyone who wants to do it with us. 🇬🇧

Agree? Sign here 👇

https://t.co/2vJCZva9Xg

See 1 related tweets

@_rockt: Signed\n\nQT @StartupCltn: 📣 Today with @Founders_Forum we've published an open letter to Andy Burnh...

13. beatsinbrief (Group Score: 66.1 | Individual: 34.1)

Cluster: 2 tweets | Engagement: 202 (Avg: 332) | Type: Tech

🚨 EXPLAINED: How Gautam Adani built one of 🇮🇳 India's largest business empires

• Began with a small commodity trading company in 1988, which later became the Adani Group.

• Focused on infrastructure, expanding into ports, logistics, power, airports, cement, and renewable energy.

• Built an integrated business ecosystem where each sector supports the others, improving efficiency and growth.

• Scaled rapidly through strategic acquisitions, including Ambuja Cements, ACC, and NDTV.

• Expanded operations across multiple international markets while strengthening India's infrastructure.

• In FY2026, the Group reported ₹94,834 crore EBITDA and invested ₹1.53 lakh crore in capital expenditure.

• Today, the Group is investing in green hydrogen, AI-ready data centres, renewable energy, and proposed nuclear power to drive future growth.

See 1 related tweets

@_TheTathya: From A Small Trading Firm To A Global Infrastructure Powerhouse: The Adani Story 🔥

Founded by Gauta...

14. chddaniel (Group Score: 65.0 | Individual: 33.1)

Cluster: 2 tweets | Engagement: 6 (Avg: 8) | Type: Tech

Apple: 'We’re removing your app from App Store'

Shipper: 'Cool, we’ll just let people mail an AI and build apps in Mail App instead'

Bro got trolled in 16K. https://t.co/1Rl0asxrC8\n\nQT @shipper_now: BREAKING: Apple is scared of vibecoders

they removed Shipper from the App Store so we migrated business building to Mail

good luck removing this one, Apple https://t.co/8TXdNWXdEk

See 1 related tweets

@chhddavid: does anyone even remember tim cook vs fortnite https://t.co/nPhLMtYRCn\n\nQT @shipper_now: BREAKING:...

15. teslaownersSV (Group Score: 65.0 | Individual: 15.5)

Cluster: 5 tweets | Engagement: 138 (Avg: 252) | Type: Tech

Grok 4.5 hits #2 on the APEX-SWE leaderboard at 51.2% Pass@1 on real-world software engineering tasks.

Leads Integration (65%) and shows a 30+ point jump from Grok 4 in one year.

Strong progress on agentic coding workflows. Congrats to the xAI team! 🚀 https://t.co/wWbGjgegKA\n\nQT @elonmusk: Grok places second after Fable on real-world software engineering

See 4 related tweets

@elonmusk: Grok 4.5 even ranks slightly above Fable, which is an incredibly good model, on some software benchm...
@Kyrannio: Really cool breakdown here of Grok 4.5\n\nQT @Xbow: Our AI research team has been analyzing affordab...
@XFreeze: RT @XFreeze: Grok 4.5 now ranks as the #1 model on AutomationBench-AA for real-world AI tasks

Scori...

@sairahul1: RT @EntelligenceAI: Grok 4.5 + Grok Build just hit #1 on SWE-Atlas-QnA with a score of 84.

That put...

16. VaibhavSisinty (Group Score: 63.5 | Individual: 63.5)

Cluster: 1 tweets | Engagement: 664 (Avg: 82) | Type: Tech

Satya Nadella just warned every company using AI: you are paying twice. Once with money. Again with something far more valuable.

He published an article introducing something called the Reverse Information Paradox.

And it changes how you think about every AI tool your company uses.

Nobel laureate Kenneth Arrow described the original paradox: a seller risks giving away knowledge just to sell it. Nadella says AI flips this completely.

In the AI age, the buyer gives away knowledge just to use what they bought.

Every time your team uses Claude or GPT at work, every prompt reveals what you are building. Every correction teaches the model what good looks like inside your company. Every eval shows what you value. Every trace exposes your workflow.

The model provider learns more about you with every interaction. You learn almost nothing about what they are learning in return. Your corrections are distilled institutional know-how. The kind a competitor could never buy.

And it leaks trace by trace, correction by correction, without you noticing. His line: "You can offload a task. You can offload a job. But you can never offload your learning."

If the model provider disappears tomorrow, do you still own the intelligence your team built on top of it? Your evals. Your memory. Your traces. Your workflows. Or did all of that compound inside someone else's infrastructure?

In the cloud era, companies accumulated data. In the AI era, they accumulate learning. Right now, most of that learning is compounding inside the model provider. Not inside the company paying for it.

The CEO pushing AI harder than anyone just told you to protect your knowledge from the very tools he is selling you. That should tell you everything.\n\nQT @satyanadella: https://t.co/xv6csf1SbV

17. alexcooldev (Group Score: 63.4 | Individual: 28.8)

Cluster: 3 tweets | Engagement: 45 (Avg: 130) | Type: Tech

The first mobile game I’ve seen built with Claude Fable 5 is now live on the App Store. Congrats, @adamlyttleapps !

I’ll be releasing my first game soon too, right after I launch my gamified workout tracker app. 💪\n\nQT @adamlyttleapps: The first Fable 5 vibe coded game is now LIVE on the app store: https://t.co/1QN6Rs8M3y

See 2 related tweets

@adamlyttleapps: The longest wait was the App Store approval

The game was ready days ago\n\nQT @adamlyttleapps: The ...

@adamlyttleapps: Sneaky people have already spotted my new secret distribution platform hiding in the game 😅\n\nQT @a...

18. rewind02 (Group Score: 62.2 | Individual: 36.8)

Cluster: 2 tweets | Engagement: 50 (Avg: 31) | Type: Tech

OpenAI finance team:

"This used to take so much manual work

Now in one pass, ChatGPT can run the variance analysis for me."

in a 34-minute launch stream, OpenAI ships GPT-5.6 and shows it running their entire company: finance, design, farming, all of it

here's what they showed:

a finance team beating forecast by $2M, then having the model rebuild the model and the deck automatically
a new desktop app that reads your files, your browser tabs, and controls other apps on your Mac
a full 3D interactive prototype built from a single prompt, no code
5.6 Soul autonomously training the next model in the family, with almost no researcher involved

the wildest claim: experts say it's better than anything else on the market while being three times as fast

a step-by-step breakdown of what actually shipped👇\n\nQT @0x_kaize: https://t.co/zE4DigQzVh

See 1 related tweets

@rewind02: RT @rewind02: OpenAI finance team:

"This used to take so much manual work

Now in one pass, ChatGPT...

19. 0x_kaize (Group Score: 61.7 | Individual: 35.6)

Cluster: 2 tweets | Engagement: 100 (Avg: 102) | Type: Tech

Just in July alone, this much has dropped:

Fable 5 - the smartest AI model in history Opus 5 - Fable-level intelligence at an accessible price Grok 4.5 - Opus-class performance, but cheaper. Elon is fully back GPT-5.6 - cheap and blazing fast. beats Fable on some benchmarks GPT-6 - Fable-level performance without worrying about limits ChatGPT Work - regular people can now get Claude Code-level agents inside ChatGPT

30 days from now your workflow will look nothing like it does today.

Dramatically less effort & overwhelmingly higher output.

And almost all of these models go live on Creao AI before anywhere else, the team tracks every release and ships new models into the product within a day.\n\nQT @CreaoAI: GPT-5.6 Sol, Terra, and Luna are live on CREAO.

Three tiers, one family: Sol for the hardest problems, Terra for balanced everyday work, Luna for fast, high-volume work at the lowest cost.

Fable 5 or GPT-5.6? Don't take our word for it. Try both and see which one earns your workload. Same platform, same infrastructure, same context and memory.

Select either. Compare for yourself.

See 1 related tweets

@0x_kaize: RT @0x_kaize: Just in July alone, this much has dropped:

Fable 5 - the smartest AI model in histo...

20. rewind02 (Group Score: 60.9 | Individual: 34.4)

Cluster: 2 tweets | Engagement: 39 (Avg: 31) | Type: Tech

Apple Silicon reviewer, live investigation:

"So basically, if you're running Apple's built-in FM tool, a Mac Mini gives you the same exact experience as the maxed out Mac Studio."

in an 11-minute deep dive, he uncovers a hidden AI built directly into Mac OS 27 that almost nobody knows exists

here's what he found:

a full LLM sitting inside the terminal, no install, no Ollama, no LM Studio needed
a $10,000 Mac Studio performing identically to a$ 4,000 Mac Mini on this tool
every monitoring tool on the market completely missing where the compute actually happens
forcing two chips to fight each other just to prove which one was secretly doing the work

the twist: the $6,000 you'd spend upgrading buys you literally nothing for this feature

a guide to choosing local AI hardware 👇\n\nQT @antisadh: https://t.co/LbXKY5PRnC

See 1 related tweets

@rewind02: RT @rewind02: Apple Silicon reviewer, live investigation:

"So basically, if you're running Apple's ...