科技热门推文——2026年7月28日

今日科技领域，开放式人工智能成为焦点。英伟达、微软及行业合作伙伴积极倡导共享模型与安全工具；与此同时，月之暗面发布参数规模达2.8万亿的Kimi K3模型，该模型具备视觉能力，并支持百万Token上下文窗口。据报道，英伟达还支持了Safe Superintelligence的大规模算力扩张计划。

在工程与市场方面，中国存储芯片制造商长鑫存储高调登陆公开市场，巨额人工智能基础设施融资也备受关注。与此同时，网络安全、平台控制权高度集中，以及为人工智能训练而破坏性收购珍稀图书等问题引发的担忧持续升温。

1. Cadence (Group Score: 983.4 | Individual: 58.5)

Cluster: 44 tweets | Engagement: 1736 (Avg: 143) | Type: Tech

RT @nvidia: AI security advances when the industry builds in the open, together.

We're introducing the Open Secure AI Alliance with industry leaders to develop new techniques and tools to safeguard software and agents.

By sharing models, tooling and research in the open, we can broaden the community of defenders.

Learn more about the founding members’ contributions: https://t.co/A16oqxs5Ty

See 43 related tweets

@steipete: We do all the amazing security work with some of the best teams in the world and people still think ...
@Salesforce: Salesforce customers want the flexibility to choose the models that work best for them—open or close...
@kimmonismus: Jensen Huang second post ever: he isn’t just arguing for open AI anymore but building an alliance ar...
@MichaelDell: Proud to join @nvidia as a founding member of the Open Secure AI Alliance. Defenders need open tools...
@SawyerMerritt: SpaceXAI is in the group to help form a new alliance to build and share open tools that promote resp...

2. CSProfKGD (Group Score: 867.1 | Individual: 64.7)

Cluster: 24 tweets | Engagement: 10040 (Avg: 799) | Type: Tech

RT @Kimi_Moonshot: Releasing the model weights and technical report of Kimi K3.

Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.

New model architecture: 2.5x the intelligence per unit of compute, not just more params.

Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale.

Model weights: https://t.co/7m7eEg6Y0B Tech report: https://t.co/yeu6cjpMCT Tech blog: https://t.co/YTfiMSNM1f

See 23 related tweets

@eliebakouch: this scaling law is a piece of art, kimi K3 recipe improves by ~2.5x over kimi K2 recipe

the tech r...

@eliebakouch: wow with kimi K3 open weights, moonshot is also open sourcing a significative improvement over deeps...
@testingcatalog: MOONSHOT 🔥: Kimi K3, a 2.8T MoE model, is now available on HuggingFace!

Kimi K3 is the first ope...

@dbreunig: Things I've already booked marked in here:
Using a knowledge graph to guide realistic task synthe...
@mark_k: The strongest AI model in the world in now open source:

Kimi K3 by @Kimi_Moonshot\n\nQT @Kimi_Moons...

3. HedgieMarkets (Group Score: 368.3 | Individual: 68.0)

Cluster: 6 tweets | Engagement: 16736 (Avg: 1566) | Type: Tech

🦔AI companies are bulk-buying rare books, scanning them through high-speed machines that cut the spines off, and shredding the originals. A service called ISBNdb facilitates orders of up to a million books and keeps buyers anonymous. Pre-2022 books are premium because they're free of AI-generated text. A federal judge ruled the practice is fair use because eliminating the original means only one copy exists at a time. Anthropic hired the former head of Google Books partnerships to obtain "all the books in the world."

My Take This got to me. A bookseller told 404 Media that rare books with almost no surviving copies are being fed into this pipeline. Books that survived wars, fires, and centuries of handling are being shredded so an AI can learn to write a better marketing email. ISBNdb's website literally says "'AI company destroys two million books' is not a headline that generates sympathy," and they still built an entire business around making it happen quietly. They offer NDAs as a feature. They coach clients to call it "digital preservation."

I've covered AI companies scraping the internet, torrenting libraries, and stealing music. This is worse because it's irreversible. You can re-upload a website. You can reprint a bestseller. You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal. So it's going to accelerate. "We shred rare books and offer NDAs so nobody finds out" is a legitimate business model in 2026. What a timeline.

Hedgie🤗

See 5 related tweets

@BrianRoemmele: Some folks don’t believe that a “principled AI company” would destroy books after they suck out all ...
@BrianRoemmele: WHAT I HAVE BEEN WARNING FOR DECADES IS HERE, THANK YOU FOR YOUR BOOK BURNING TOO ANTHROOIC.

“Books...

@BrianRoemmele: Confirmation of Anthropic’s BOOK BURNING for their AI: The Great Forgetting of the Amnesia Generatio...
@BrianRoemmele: “This could never happen again, we won’t stand for book burning again…”—2026

It already is happenin...

@danshipper: if true, this sucks and shouldn't continue with rare books\n\nQT @HedgieMarkets: 🦔AI companies are b...

4. IPONewsroom_ (Group Score: 349.0 | Individual: 29.8)

Cluster: 17 tweets | Engagement: 66 (Avg: 19) | Type: Tech

CXMT is now a public company ... CXMT is the largest Chinese maker of memory chips

CXMT was up 466% in its first day of trading as a public company in China 🇨🇳 and is now worth around $500 Billion

CXMT is now one of the largest public Chinese companies https://t.co/r2GejiQAyR\n\nQT @KraneShares: In a recent @barronsonline article, @luoshanji covered China's two memory chip giants going public, ChangXin Memory Technologies (CXMT) and Yangtze Memory Technologies (YMTC), at a moment when AI-driven demand has created a global memory shortage and sent prices soaring.

The piece highlighted the KraneShares China Technology & Semiconductor STAR Market 50 Index ETF (Ticker: $KSTR) for investors looking to access CXMT, noting that its size makes it a plausible future constituent of the STAR Market 50 index that KSTR tracks, with CXMT's first-quarter revenue surging 719% year over year and its IPO set to value the company at roughly$ 85 billion when it begins trading July 27.

Read: https://t.co/MjE6FBvtzl

For KSTR standard performance, top ten holdings, and other fund information: https://t.co/P4lnYtj6iZ

See 16 related tweets

@WOLF_Financial: Chinese memory chip maker CXMT is going public tomorrow in Shanghai, China 🇨🇳

CXMT is the 2nd larg...

@MilkRoadAI: RT @MelvinInvests: China's newest chip star just IPO'd up over 450% and it's bullish for memory (Sav...
@danielnewmanUV: RT @rolfbulk: Major IPO in China today, with $CXMT becoming the most valuable company listed on main...
@IPONewsroom_: CHINA'S LARGEST MEMORY CHIPMAKER JUST POPPED ABOUT 470% IN ITS SHANGHAI DEBUT

CXMT priced its IPO ...

@business: CXMT, a secretive Chinese chipmaker, has made its trading debut after a $9.8 billion IPO.

It's set ...

5. rohanpaul_ai (Group Score: 320.1 | Individual: 26.0)

Cluster: 15 tweets | Engagement: 60 (Avg: 39) | Type: Tech

Ilya Sutskever’s Safe Superintelligence partners with Nvidia to scale its compute 10X.

NVIDIA also invested in SSI. While they did not disclose the investment amount, Financial Times reported its a $5B commitment.

The latest partnership adds an NVIDIA GPU path to SSI’s earlier Google Cloud TPU arrangement.

SSI says its research is now worth scaling after 2 quiet years, yet it has released no product and published no research.

SSI’s latest publicly available valuation is $32B post-money, set during its April 2025 funding round.\n\nQT @ssi: We are announcing a long-term strategic partnership with NVIDIA. NVIDIA is making a substantial investment in SSI that will let us 10x our compute in the next 12 months. We reached the point where our research is worth scaling and with this partnership we will be able to. We are honored by NVIDIA’s conviction.

See 14 related tweets

@WSJ: Exclusive: Nvidia is investing in former OpenAI Chief Scientist Ilya Sutskever’s secretive artificia...
@mark_k: Ilya Sutskever's @ssi is partnering with Nvidia.

"We've reached a point where our research is worth...

@teortaxesTex: So that, too, is the "closed models" that the open models Coalition supports. I've just recently th...
@StockMKTNewz: NVIDIA JUST INVESTED IN ILYA SUTSKEVER'S SAFE SUPERINTELLIGENCE

Safe Superintelligence Inc. (SSI) a...

@firstadopter: SSI is no longer a TPU only shop.

NVIDIA: "Safe Superintelligence Inc. (SSI) and NVIDIA today annou...

6. novita_labs (Group Score: 315.9 | Individual: 57.5)

Cluster: 10 tweets | Engagement: 58980 (Avg: 4569) | Type: Tech

RT @JensenHuang: For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

See 9 related tweets

@UnslothAI: We signed the Open Weights letter because we believe the future of AI should be shaped by everyone, ...
@WesRoth: Jensen Huang just made his first-ever post on X.

And he used it to defend open AI.

Every company r...

@XFreeze: SpaceX has officially signed the letter backing one of the strongest industry-wide declarations in s...
@CNBC: Nvidia and a host of tech giants on Monday launched a new artificial intelligence safety initiative ...
@glean: Proud to join the coalition supporting a more open, competitive AI ecosystem.

A more open ecosystem...

7. amasad (Group Score: 232.6 | Individual: 40.6)

Cluster: 7 tweets | Engagement: 566 (Avg: 244) | Type: Tech

Interesting drop from former Anthropic employee: Hackers prefer to use massively subsidized labs AI subscriptions for attacks as opposed to open models. https://t.co/gSVw9tv346\n\nQT @NoahLebovic: I don't think it's a lack of imagination. I also used to work at Anthropic, think trends will continue, and used to agree with this. But I've changed my mind and now disagree with this take.

Open models are already capable enough to do what you described. For example, I used Opus 4.6 to gain access to other folks medical records, hijack bank accounts, etc. back in February. GLM 5.1 is more capable than Opus 4.6 in most pentesting environments, and it came out in April.

Despite capable open-weight models existing, the sketchier folks I know are still using a Claude Code or Codex subscription for hacking. (Even well-resourced groups in other countries! They use the grey/black market of discounted Ant/OAI subscription tokens sold through resellers.) So I see most of the materialized risk here as still coming from Anthropic and OpenAI; safeguards aren't sufficient to stop a moderately dedicated actor.

The groups I know who are using open-weight models are legitimate offensive security companies. They won't break the rules to use subscription-based pricing, the open-weight models are more reliable in that they don't require specific jailbreaks nor hit classifiers, and the labs use massive partnerships or spend as a prereq for lowering classifiers/safeguards. I know of three legitimate groups running GLM 5.2 as their primary model.

That last part applies for Anthropic, too: I know of two instances where two different Anthropic GTM people used large comitted spend contracts as a prereq for lowering safeguards, and I directly witnessed one.

On the inside, I know the narrative and intent is genuinely about safety. But from the outside, Anthropic-the-system seems to be optimizing for revenue and control/power, isn't diffusing capabilities to defenders, and also doesn't have adequate safeguards to prevent misuse from dedicated bad actors.

As a result, I now lean towards a future where capable open models are freely available (at least for cyber, bio is harder); I don't trust Anthropic or other frontier labs to handle this sufficiently well without diffused capabilties given what I've seen so far.

See 6 related tweets

@teortaxesTex: Fascinating Thanks Noah I think everyone will eventually accept Teortaxes Thought. We just still ne...
@BrianRoemmele: Wow!

Former Anthropic engineer has made claims that the company lowered guardrails for big custome...

@zephyr_z9: Well...\n\nQT @NoahLebovic: I don't think it's a lack of imagination. I also used to work at Anthrop...
@Thom_Wolf: 👀\n\nQT @NoahLebovic: I don't think it's a lack of imagination. I also used to work at Anthropic, th...
@zephyr_z9: RT @NoahLebovic: I don't think it's a lack of imagination. I also used to work at Anthropic, think t...

8. business (Group Score: 173.6 | Individual: 30.2)

Cluster: 7 tweets | Engagement: 92 (Avg: 54) | Type: Tech

Moonshot AI is poised to make its Kimi K3 model available for public download, expanding its reach in the global open software community at a time of growing US concern about Chinese encroachment into the top echelons of AI development https://t.co/zIOoU0lXoR

See 6 related tweets

@Kimi_Moonshot: Kimi K3 is now available on @digitalocean 's Serverless Inference!
Developers can start building w...
@vllm_project: Happy to collaborate with @digitalocean to run Kimi K3 on vLLM! Great to see it live. 🚀\n\nQT @digi...
@business: Moonshot AI is poised to make its Kimi K3 model available for public download, expanding its reach a...
@Cointelegraph: 🇨🇳 JUST IN: China's Moonshot AI is set to release its Kimi K3 model weights for free public download...
@business: Moonshot AI has made its Kimi K3 model available for public download. Here's how that could change t...

9. Analyticsindiam (Group Score: 145.9 | Individual: 56.6)

Cluster: 4 tweets | Engagement: 3122 (Avg: 213) | Type: Tech

RT @PawanKChandana: Honoured to meet Hon’ble Prime Minister Shri @narendramodi ji along with my co-founder, Bharath.

From two engineers who started out with a bold idea which had low chance of success—to standing together with the Prime Minister after Vikram-1 became India’s first private orbital launch vehicle to reach orbit, it was a deeply humbling moment.

Your vision and bold reforms opened India’s space sector to private enterprise and empowered a new generation of innovators. We are proud that @SkyrootA could contribute to this national journey.

Committed to building from India, for the world. 🇮🇳🚀

See 3 related tweets

@SkyrootA: An honour to meet Hon’ble Prime Minister Shri @narendramodi.

Our founders, Pawan Chandana and Naga ...

@narendramodi: A great interaction with Pawan Chandana and Naga Bharath Daka of Skyroot Aerospace.

The success of...

@mygovindia: Inspiring India's Next Space Revolution!

The meeting between the @SkyrootA team and PM @narendramod...

10. cryptopunk7213 (Group Score: 135.2 | Individual: 35.7)

Cluster: 6 tweets | Engagement: 114 (Avg: 78) | Type: Tech

just to recap - china’s had an insane week in ai:

kimi k3 open model at frontier intelligence
cxmt (ai memory maker) IPO’d and up 500% valuing them at $500B (most valuable company in china)
news broke this morning they’ve developed a key component of ASML’s tech that is required to make ai chips

probably just a coincidence but they’ve really gained momentum\n\nQT @KobeissiLetter: AI demand is becoming a key driver of China's economy:

China's chip exports are up to a record ~$290 billion over the last 12 months.

This figure has more than doubled over the last 3 years.

At the same time, computer and component shipments are up to ~$240 billion, the highest since 2022.

Power equipment exports are up to a record ~$190 billion, with the growth accelerating over the last 2 years.

This brings China’s total AI-related exports up to ~$720 billion over the last 12 months, an all-time high.

Western demand for China-made AI hardware is surging.

See 5 related tweets

@MelvinInvests: RT @MelvinInvests: China just started mass producing its own chipmaking tools and it's a big milesto...
@KobeissiLetter: AI demand is becoming a key driver of China's economy:

China's chip exports are up to a record ~$29...

@StockSavvyShay: $CXMT has surpassed a$ 500B market cap becoming China’s most valuable publicly traded company.

AI-d...

@azeem: In 2015, Made in China set a target of 70% self-sufficiency in semiconductors within a decade. It wa...
@Cointelegraph: 🇨🇳 BIG: AI demand pushes China's AI-related exports to a record ~$720 billion over the past 12 month...

11. cryptopunk7213 (Group Score: 124.9 | Individual: 29.1)

Cluster: 6 tweets | Engagement: 190 (Avg: 78) | Type: Tech

this is the single biggest data center deal ever

nvidia’s committing $600 billion to fund 10GW of compute for OpenAI. thats 2X nvidia’s entire annual revenue lol

largest financing loan by 2-3x. howard lutnick (US commerce sec.) is personally involved.

japan financing the power.\n\nQT @KobeissiLetter: BREAKING: Nvidia is in talks with OpenAI to guarantee $250 billion in financing for a data center in Ohio.

Details include:

The guarantees from Nvidia would help OpenAI lease a 10GW project that SoftBank is developing in Ohio
In total, the project could cost more than $500 billion, the largest data center project ever announced
The power for the project is controlled by the US government and funded separately by Japan under a recent trade deal
Commerce Secretary Lutnick is reportedly involved in deciding who will get the power

The data center buildout just hit a whole new level.

See 5 related tweets

@HedgieMarkets: 🦔The Wall Street Journal reported today that Nvidia is in talks to guarantee about $250 billion in d...
@danielnewmanUV: $NVDA backstopping OpenAI’s Ohio project is pretty damn bullish. Probably even helped Sam become an...
@Reuters: Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports https:/...
@CNBC: Nvidia and OpenAI in talks for up to $250 billion dollar backstop to fund AI infrastructure plans ht...
@Forbes: Nvidia And OpenAI Discussing $500 Billion Data Center—Here’s What We Know https://t.co/TPPGcftgde ht...

12. garrytan (Group Score: 119.5 | Individual: 45.4)

Cluster: 5 tweets | Engagement: 6394 (Avg: 284) | Type: Tech

RT @JensenHuang: Attackers have frontier AI. Defenders need a frontier AI ecosystem—the best open and closed models, force-multiplied by a global community.

During the Hugging Face incident, closed AI blocked essential forensics. An open-weight frontier model helped contain the intrusion.

That’s why we created the Open Secure AI Alliance.

See 4 related tweets

@tunguz: The only way to stop a bad guy with an LLM is with a good guy with an LLM.\n\nQT @JensenHuang: Attac...
@elonmusk: Extremely important point\n\nQT @JensenHuang: Attackers have frontier AI. Defenders need a frontier ...
@tetsuoai: It's growing. https://t.co/ng22gNM2DN\n\nQT @JensenHuang: Attackers have frontier AI. Defenders need...
@edzitron: Very desperate for him to be posting this much\n\nQT @JensenHuang: Attackers have frontier AI. Defen...

13. WesRoth (Group Score: 110.9 | Individual: 36.1)

Cluster: 6 tweets | Engagement: 84 (Avg: 32) | Type: Tech

Satya Nadella just entered the fight over who gets to control AI.

Microsoft says open-weight models are essential to American leadership.

The argument is simple:

If only a few companies control the most capable models, they also control the price, access, infrastructure, and direction of the AI economy.

Open weights give developers and businesses the ability to download models, modify them, run them privately, and build without depending on one provider.

Yes, that creates security risks.

But Microsoft believes concentrating AI inside a few closed systems may create an even bigger risk.

This is a battle over whether the future of AI belongs to a small group of labs or to the entire ecosystem building on top of them.\n\nQT @satyanadella: Open-weight models are essential to a healthy AI ecosystem. Together with others across our industry, we are outlining a path for open-weight models to strengthen American competitiveness and expand economic opportunity, while protecting national security. https://t.co/Tr0sAzAxTD https://t.co/opMwEbtpoq

See 5 related tweets

@haider1: did you know?

Anthropic comes from the Greek word "ánthrōpos", meaning "human being"

and yet the c...

@WesRoth: Google just picked a side in the open AI fight.

Google is officially supporting the industry push t...

@haider1: open-weight models directly threaten Anthropic leadership's core beliefs

they want to control who g...

@BrianRoemmele: RT @UnslothAI: We signed the Open Weights letter because we believe the future of AI should be shape...
@PrismML: RT @BabakHassibi: At PrismML we believe an open-weight AI ecosystem is critical to fostering innovat...

14. Kimi_Moonshot (Group Score: 106.8 | Individual: 30.6)

Cluster: 4 tweets | Engagement: 602 (Avg: 1356) | Type: Tech

Happy to have @FireworksAI_HQ as our day0 launch partner and bring Kimi K3 to more developers.

With Fireworks, you can now deploy and fine-tune the 2.8T Kimi K3 model with just a few clicks! https://t.co/mwQKjvXGdl\n\nQT @FireworksAI_HQ: Kimi K3 is live on Fireworks. Day 0, inference and training. US-hosted, and zero data retention.

This is the first frontier open model in the 3 trillion parameter class. It sports 1M context, native vision, and reasoning that rivals the top closed models.

Boom. https://t.co/ItU3KAKt61

See 3 related tweets

@omarsar0: Kimi K3 is now available for inference & training in @FireworksAI_HQ.

Crazy how easy they make ...

@Kimi_Moonshot: Excited to have @modal as Day 0 launch partner for Kimi K3!

They trained a custom DFlash speculator...

@dair_ai: RT @omarsar0: Kimi K3 is now available for inference & training in @FireworksAI_HQ.

Crazy how e...

15. BrianRoemmele (Group Score: 105.8 | Individual: 41.2)

Cluster: 3 tweets | Engagement: 469 (Avg: 285) | Type: Tech

The Robber Barons Built Cathedrals of Knowledge. Anthropic Built a Shredder.

History remembers the robber barons with a peculiar mixture of contempt and reluctant gratitude. They crushed competitors, broke unions, polluted rivers, and extracted fortunes that would make modern billionaires blush.

Yet Andrew Carnegie, the most ruthless of them in steel, spent the final decades of his life doing something almost unrecognizable in today’s technology culture: he built public libraries.

Nearly 2,500 of them. One thousand six hundred eighty-one in the United States alone. He poured the equivalent of well over a billion dollars in today’s money into brick, mortar, oak shelves, and free books so that any working man or woman could walk through the door and claim a share of human knowledge.

Carnegie did not do this out of soft-hearted liberalism. He believed that the free library was the single highest form of philanthropy because it equipped people to lift themselves. The buildings still stand.

The books still circulate. The knowledge was not merely copied into some private vault; it was made permanently available to the public as a shared inheritance.

Now look at Anthropic.

In early 2024 the company launched an internal effort it code-named Project Panama. An unsealed planning document put the ambition in language so stark it reads like a confession: “Project Panama is our effort to destructively scan all the books in the world.” The same document warned employees, “We don’t want it to be known that we are working on this.”

I caution you… what you read may get you mad.

Ready?

Read on below…\n\nQT @BrianRoemmele: https://t.co/FAYkOUohnN

See 2 related tweets

@BrianRoemmele: Meet Mr. Wordsworth.

He was declared obsolete by the state.

Why? The state and companies eliminate...

@BrianRoemmele: RT @BrianRoemmele: The Robber Barons Built Cathedrals of Knowledge. Anthropic Built a Shredder.

His...

16. lennysan (Group Score: 103.5 | Individual: 36.3)

Cluster: 3 tweets | Engagement: 131 (Avg: 114) | Type: Tech

Internal team principles from Anthropic's AI research product team https://t.co/Xzo5VFpB44\n\nQT @lennysan: Dianne Penn is Head of Product for Research and Labs at @AnthropicAI. She joined in 2023 as their first technical PM (when there were just five product engineers) and has helped ship every Claude model from Claude 2 to Mythos.

We discuss: 02:31 Early Anthropic 08:55 The two biggest inflections 13:50 Inside the exponential 20:02 Token maxing 23:30 Anthropic Labs and the incubation model 27:30 How the AI research role works 31:35 How to become a top AI researcher 35:18 Frontier model safeguards 39:38 Hiring in the AI era 44:16 Building evals 47:48 Evals are the new PRDs 49:55 The importance of hands-on leadership 52:46 Finding joy in AI 58:10 How Dianne uses Claude 01:01:05 Avoiding overreliance on AI 01:03:50 The constitution that makes Claude better 01:07:11 AI writing and verification 01:11:40 Where human brains will continue to be valuable 01:14:10 Navigating AI with kids 01:16:26 Alignment, the future of the PM role, and burnout 01:21:54 Lightning round and final thoughts

See 2 related tweets

@lennysan: How @AnthropicAI's Head of Product for AI research avoids AI brainrot https://t.co/reSg3iebmo\n\nQT ...
@lennysan: "You need frontier products in order to have frontier models and for people to feel the magic of fro...

17. XFreeze (Group Score: 102.6 | Individual: 36.6)

Cluster: 4 tweets | Engagement: 661 (Avg: 410) | Type: Tech

Grok 4.5 just got even cheaper to run

As Elon promised, SpaceXAI has cut cached-input pricing from $0.50 to$ 0.30 per million tokens......a 40% reduction. For contexts above 200K tokens, cached input dropped from $1.00 to$ 0.60

This matters a lot for: • Long-running coding agents • Multi-turn conversations • Repeated repository context • Tool-heavy workflows • Large system prompts • Autonomous agent loops

Successful cache hits now cost nearly 7× less than regular input. At 100 million cached tokens, the cost falls from $50 to$ 30......pure savings that scale fast on real production workloads

Grok 4.5 was already fast and token-efficient. Now it’s even more cost-efficient to build with\n\nQT @XFreeze: Grok 4.5 is now live in the SpaceXAI API

Pricing: • $2/M input tokens •$ 0.50/M cached input tokens • $6/M output tokens

Fast, cost-efficient, and built for coding, agents, engineering, and knowledge work

• 500K context window • 80 tokens/sec • Tool calling • Advanced reasoning • Structured outputs • Vision support • 150 requests/sec • 50M tokens/min

It also supports API rate limits of up to 7,200 RPM, 150 RPS, and 50M TPM (depending on your assigned limits/tier)

See 3 related tweets

@XFreeze: Grok Build with Grok 4.5 stands far ahead on the efficiency frontier....completely alone inside the ...
@elonmusk: RT @XFreeze: Grok Build with Grok 4.5 stands far ahead on the efficiency frontier....completely alon...
@RoundtableSpace: GROK 4.5 JUST CUT CACHED-INPUT PRICING BY 40%, MAKING LONG-RUNNING AI AGENTS FAR CHEAPER TO OPERATE....

18. Cadence (Group Score: 100.2 | Individual: 18.4)

Cluster: 7 tweets | Engagement: 9 (Avg: 143) | Type: Tech

Exciting news with @nvidia! 🤝 #DAC2026\n\nQT @NVIDIAAIInfra: 📣 Announcing the expansion of the NVIDIA Agent Toolkit.

At #DAC2026, we've added NVIDIA PhysicsNeMo and NVIDIA CUDA-X libraries, enabling developers to build customizable AI agents for electronic and systems design. 🎉

Industry leaders including @Cadence, @Siemens, @Synopsys and more are adopting these capabilities to advance chip-to-systems designs workflows.

Learn more ➡️ https://t.co/tqlDgtWurK

See 6 related tweets

@Synopsys: Live updates from #DAC2026. Excited to be part of today's news. 🤝 @NVIDIAAIInfra\n\nQT @NVIDIAAIInfr...
@Cadence: RT @NVIDIAAIInfra: 📣 Announcing the expansion of the NVIDIA Agent Toolkit.

At #DAC2026, we've adde...

@Cointelegraph: 🔥 BIG: Nvidia expands its Agent Toolkit with new PhysicsNeMo and CUDA-X libraries, giving developers...
@FirstSquawk: NVIDIA enhances its Agent Toolkit with PhysicsNeMo and CUDA-X libraries to transform how industries ...
@financialjuice: Nvidia expands agent toolkit with PhysicsNemo and CUDA-X libraries to transform global engineering, ...

19. aakashgupta (Group Score: 96.5 | Individual: 29.1)

Cluster: 4 tweets | Engagement: 179 (Avg: 165) | Type: Tech

The open weights letter launched Thursday with 25 signatures. Reports put it near 50 a day later, and names keep landing.

Bolt's is the one I'd flag. Eric's post below lays out the record: years of funding the OSS projects the web runs on, the OSS Pledge, open sourcing https://t.co/va8BifidIC. And Bolt sits at the application layer, where open weights are supply chain. What stays downloadable, inspectable, and swappable underneath decides what you can ship.

The range of the list is the signal. Chip makers, clouds, labs, VCs, and a growing slice of the application layer are signing the same page, which makes it harder to wave off as labs talking their own book.

Every AI product carries assumptions about what stays open underneath it. Bolt just put its assumptions in writing.\n\nQT @EricSimons: Over the weekend Bolt signed the Open Weights and American AI Leadership letter.

For years we've put real money behind OSS: funding the projects the web runs on, joining the OSS Pledge, open sourcing @boltdotnew & more.

Open weights are the logical next chapter of that same story. https://t.co/VdqeXt5B3V

See 3 related tweets

@rohanpaul_ai: Love the direction here and so great to see Bolt joining the letter.

@boltdotnew has a history of f...

@boltdotnew: RT @EricSimons: Over the weekend Bolt signed the Open Weights and American AI Leadership letter.

Fo...

@boltdotnew: RT @aakashgupta: The open weights letter launched Thursday with 25 signatures. Reports put it near 5...

20. rohanpaul_ai (Group Score: 96.0 | Individual: 36.8)

Cluster: 3 tweets | Engagement: 94 (Avg: 39) | Type: Tech

Nebius Token Factory just brought Kimi K3 as an OpenAI-compatible API.

World's first open-weight model to reach frontier-level performance.
up to 1M tokens context window
2.8T-parameter Mixture-of-Experts architecture that activates only 16 of 896 experts per token.
On the Intelligence Index, Kimi K3 scores 57, only two points below GPT-5.6 Sol.\n\nQT @nebiustf: Kimi K3 is now available on Token Factory.

We’re excited to announce that Nebius Token Factory is an official Day 0 partner for @Kimi_Moonshot's Kimi K3.

Kimi K3 is the first open-weight model to reach frontier-level performance, a major step forward for open models.

It is built for long-horizon coding, knowledge work and reasoning, with native vision and up to 1M tokens of context. Artificial Analysis scores it at 57 on its Intelligence Index, just two points behind GPT-5.6 Sol (max). That puts Kimi K3 at the top of the open-weight field and firmly among today’s frontier models.

Developers can access K3 through Token Factory’s OpenAI-compatible API and console today.

Give K3 the hard problem.

Build with Kimi K3: https://t.co/olJZmAvmDh

See 2 related tweets

@ModelScope2022: 🚀 We're excited to welcome Kimi K3 to ModelScope, the world's first open 3T-class model! 🤖 Model: ht...
@ying11231: The game (made by Kimi K3) is a bit too fast in showing how many optimizations have been added to ge...