科技热门推文 - 2026-09-09

今日科技动态，AI智能体成为焦点。OpenAI声称，其智能体已求得纳维–斯托克斯方程的解；据报道，Meta正推出Muse，用于跨应用处理日常任务。Mistral宣布完成30亿欧元融资，凸显投资者的投资热情；与此同时，谷歌云与埃森哲正扩大企业级AI部署。工程技术进展涵盖机器人仿真、边缘端模型及专用尽职调查智能体，新型图像与3D工具也进一步拓展了创意应用场景。据报道，DeepMind推出的基因组图谱展现了AI在科学研究领域的雄心，但有关数学突破的说法仍有待独立验证。

1. jxnlco (Group Score: 738.2 | Individual: 46.3)

Cluster: 36 tweets | Engagement: 26318 (Avg: 1069) | Type: Tech

RT @OpenAI: We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.

The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.

The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.

See 35 related tweets

@yacineMTB: Big congratulations on this historic moment. Furthermore, I wish anthropic good luck on their upcomi...
@teortaxesTex: Пиздец. https://t.co/bS5yg3VcgU\n\nQT @OpenAI: We’re sharing a solution to the Navier-Stokes Millenn...
@AiBreakfast: Friendly reminder that the internal models are light years ahead of what they release\n\nQT @OpenAI:...
@eliebakouch: did yann lecun ever update this slide? might be a good time..

test time compute scaling laws hit ju...

@eliebakouch: i don't think we realize how early we are, the internal model by oai started training less than 2 we...

2. wallstengine (Group Score: 445.7 | Individual: 40.0)

Cluster: 21 tweets | Engagement: 753 (Avg: 179) | Type: Tech

$META LAUNCHES MUSE PERSONAL AI AGENT

Meta is rolling out Muse, a personal AI agent that can take actions across apps and the web, including sending emails, booking travel, filling out forms and making purchases with user approval.

Muse runs on Meta’s new Muse Secure VM, giving each user a dedicated cloud environment with a separate Sentinel agent that reviews actions before they reach the internet.

The agent can keep working after the app is closed, remember user preferences and proactively advance longer-term tasks. Stripe’s Link is built in for checkout, with Shop Pay and 1Password support coming later.

Muse launches in the U.S. on iOS, Android and the web, with free and paid tiers, and support for AI glasses coming soon.

See 20 related tweets

@wallstengine: $META launches Muse, its new personal AI agent that can actually take actions on your behalf. Coming...
@levie: Personal assistant agents are going to be a very exciting AI category. It’s the first time you can h...
@kimmonismus: Meta’s Muse is a personal AI agent with its own computer, browser, and a WhatsApp interface.

Accor...

@testingcatalog: META 🔥: Muse Agent has been announced!

Powered by Muse Spark 1.3 👀 Available on both iOS and A...

@spencerbarnett: Muse has hatched! 🐣

And it is truly remarkable. Browser use has been the biggest unlock for me. I ...

3. matteocollina (Group Score: 408.2 | Individual: 42.9)

Cluster: 13 tweets | Engagement: 904 (Avg: 181) | Type: Tech

RT @MistralAI: Today marks a major step for Mistral: we’re announcing a €3B Series D, the largest equity round ever raised by a European tech company, just three years after launch. https://t.co/5AexUxyrZM

See 12 related tweets

@ClementDelangue: Bravo to Mistral!

So great to see them killing it, they are an inspiration to everyone as we need ...

@testingcatalog: Mistral AI raised €3 billion in a Series D funding at more than €21 billion in a post-money valuatio...
@Thom_Wolf: Mistral raising this massive 3B round is a very good news for open-source.

Largest equity round by ...

@Rasmic: You don’t understand yet but the vibe from mistral is that they will make such a legendary comeback ...
@Gorden_Sun: Mistral宣布完成30亿欧元的D轮融资

公司估值超过210亿欧元，这也是欧洲科技公司历史上规模最大的一笔股权融资。

这轮融资由三星电子领投，多家欧洲、北美和亚洲的知名投资机构以及卢森堡大公国也...

4. svpino (Group Score: 403.1 | Individual: 37.6)

Cluster: 12 tweets | Engagement: 268 (Avg: 178) | Type: Tech

I used to work on computer vision models for robots and drones, and nothing is slower than having to run the robot in order to test your code.

This promises to solve that.

They will build a digital twin that will simulate the physical world.

Digital twins are virtual models of real physical objects or systems that use real-world data to reflect how the real physical object behaves. You can use these digital twins to test everything.\n\nQT @antiochrobotics: Today, we’re announcing Antioch’s $32 million Series A, led by @GreylockVC with participation from @A_StarVC, @Category_VC, @BoxGroup, @IcehouseVenture, and angels.

While AI has drastically accelerated software development, physical autonomy has been constrained by slow, expensive hardware-based development. Antioch enables physical AI teams to build, test, and validate systems at the speed of software.

We’re proud to be working with leading teams including @amazon @ring, @NVIDIARobotics, and @nebiusai as we make scaled, high-fidelity simulation the standard for physical AI development.

Our sincere thanks to the legion of investors, customers, partners, and advisors who are making this step change possible.

We’re scaling rapidly and hiring across simulation, infrastructure, machine learning, 3D graphics, go-to-market, and operations.

See 11 related tweets

@aakashgupta: Antioch's Series A is a bet that robots learn faster in software.

Software teams ship a change, run...

@Parul_Gautam7: What stood out to me about Antioch is the focus on iteration.

Physical AI has so many real-world co...

@omarsar0: RT @antiochrobotics: Today, we’re announcing Antioch’s $32 million Series A, led by @GreylockVC with...
@wallstengine: PHYSICAL AI STARTUP ANTIOCH RAISES $32M SERIES A LED BY GREYLOCK

Antioch is building cloud-scale si...

@StockSavvyShay: Physical AI startup Antioch just raised $32M to make testing robots as cheap and scalable as testing...

5. OtakuMachine (Group Score: 371.4 | Individual: 50.3)

Cluster: 12 tweets | Engagement: 558 (Avg: 65) | Type: Tech

RT @Buzzy_now_AI: Introducing Buzzy 3D Director Console, powered by GPT-6 Astra.

Let GPT-6 turn your words into fully interactive 3D world, and then to stunning video on Buzzy.

Build complex 3D characters & scenes with simple prompts Control cameras, characters, lighting, and environments easily Direct and refine every detail with simple prompts

Experience OpenAI's strongest model on Buzzy!

See 11 related tweets

@Lacoste0x_: AI video is becoming less about hitting “generate” and hoping for the best.

Tools like this give yo...

@OtakuMachine: This could remove a lot of friction from 3D workflows.

You describe what you want, build the scene,...

@OliviaReedai: Text → interactive 3D world → finished video.

The fact that you can keep directing and refining eve...

@EmmaUsesAi: One thing I’ve wanted from AI video tools: more control before generation.

Buzzy’s 3D Director Cons...

@SynapseOpsAI: The next step for AI video isn’t just better generation.

It’s better direction.

Being able to buil...

6. reach_vb (Group Score: 314.0 | Individual: 29.8)

Cluster: 16 tweets | Engagement: 230 (Avg: 319) | Type: Tech

Introducing ChatGPT Images 2.5 🎨

Up to 50% lower generation latency vs Images 2.0 More natural lighting, textures, and recognizable subjects Better detail preservation across repeated edits Comment directly on images to target changes Better handling of complex layouts, styles, and transparent backgrounds

Plus, new tools:

@ Sketch: draw in ChatGPT to guide your image Templates: start with posters, merch, and more Share prompts alongside images for others to remix

Two new API models, available now:

GPT-Image-2.5 Flare: better quality, editing, and speed GPT-Image-2.5 Sunburst: greater precision for detailed work, with longer generation times

Rolling out today across all ChatGPT, ChatGPT Work, and Codex tiers on desktop, mobile, and web.

Enjoy!\n\nQT @OpenAI: ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you can dream of.

Faster image generation to keep your ideas flowing
Improved fidelity for more natural, recognizable images
Consistent details across multiple edits
Comment-based edits to change only what you want

See 15 related tweets

@shiri_shh: ChatGPT Image 2.5 vs ChatGPT Image 2.0

Prompt: Create a hairstyle analysis graphic using this portr...

@CSProfKGD: RT @OpenAI: ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you...
@aiedge_: How to create amazing images with ChatGPT Images 2.5 https://t.co/yRZB9M5DDf\n\nQT @OpenAI: ChatGPT ...
@Dimillian: RT @reach_vb: Introducing ChatGPT Images 2.5 🎨

Up to 50% lower generation latency vs Images 2.0 ...

@gdb: ChatGPT Images 2.5 — another step for creating whatever you can dream of.\n\nQT @OpenAI: ChatGPT Ima...

7. CNBC (Group Score: 303.8 | Individual: 41.2)

Cluster: 15 tweets | Engagement: 132 (Avg: 59) | Type: Tech

French AI startup Mistral has raised 3 billion euros ($3.5 billion) in fresh funding led by memory chip giant Samsung.

➡️Mistral CEO Arthur Mensch told CNBC that the funds will go toward building more infrastructure, including its own data centers.

➡️The capital raising round values the company at more than 21 billion euros ($24 billion), double the level of a year ago.

🔗For more, please click here: https://t.co/MQV1Qc2taT

See 14 related tweets

@TechCrunch: The French AI lab has raised €3 billion at a €21 billion valuation in a Series D round led by Samsun...
@etnshow: JUST IN: Mistral Co-Founder and CEO Arthur Mensch believes that new fundraising gives them comparabl...
@mark_k: Huge news for European AI: @MistralAI has raised €3 BILLION at a valuation of more than €21 billion,...
@StockMKTNewz: SAMSUNG ACQUIRES EQUITY IN MISTRAL AI FOLLOWING FUNDING ROUND\n\nQT @IPONewsroom_: MISTRAL AI RAISES...
@StockMKTNewz: Mistral AI is now worth more than $24.3 Billion\n\nQT @IPONewsroom_: MISTRAL AI RAISES RECORD €3B AT...

8. MTSlive (Group Score: 301.4 | Individual: 49.2)

Cluster: 12 tweets | Engagement: 6360 (Avg: 251) | Type: Tech

SITUATION DETECTED: Google DeepMind announced AlphaGenome Atlas, a 1-petabyte map of predicted molecular effects for all 9 billion possible single-letter DNA changes in the human genome.

See 11 related tweets

@NewsFromGoogle: RT @GoogleDeepMind: We’re launching AlphaGenome Atlas: an AI-powered searchable database mapping the...
@kimmonismus: Thats really cool: Google DeepMind just released a predictive map of every possible single-letter DN...
@rseroter: Hey, that's cool. And pretty amazing.

The @GoogleDeepMind AlphaGenome Atlas platform lets you predi...

@shiri_shh: The most underrated AI release of the year just happened and barely anyone outside genomics is payin...
@etnshow: BREAKING: Google DeepMind release AlphaGenome Atlas, a free database predicting the effects of all 9...

9. teortaxesTex (Group Score: 293.2 | Individual: 38.0)

Cluster: 9 tweets | Engagement: 265 (Avg: 86) | Type: Tech

I think if you're whatsoever close to math, you're about to publish the solution to Navier-Stokes, and the thing you're worried about is the appropriateness of Anthropic-OpenAI coauthorship, you're full of shit and I don't trust anything you say. pox on both your fiefdoms. https://t.co/blhilDvzOv\n\nQT @SebastienBubeck: I would like to clarify a few things:

The screenshot is my reaching out to Levent to coordinate our releases. I hope it’s clear from the message that we came in with the best possible intentions.
I never ever asked for Levent to be removed from authorship of his own work (as indicated by my text). I was surprised to learn during the call with Tristan that they had only solved Euler and not Navier-Stokes; after learning this we brainstormed possible paths forward. One option we discussed was that Tristan could be the lead author on a rewrite of OpenAI’s Navier-Stokes proof. It is in that context that I said “it would be simpler if Levent was not an Anthropic employee” because I felt it would be inappropriate for an Anthropic employee to author OpenAI’s work. Importantly it was admitted that internal Anthropic models had been used in their proof of Euler blowup; I therefore felt I could not consider Levent to be an independent academic. Another option I wanted to propose (but got cut short) is to offer access to our internal model so that they could try to finish their proof and bridge the gap between Euler and NS. Again I did not know how to navigate giving access to internal OpenAI IP to an Anthropic employee.
To reiterate it plainly: as my text clearly indicates, and as I said during our call, OpenAI's intention was to do everything possible to celebrate their mathematical achievements and the heroic efforts that they made on Euler. In the call I was immediately met with a litany of slander, including direct threats that if we were to announce Navier-Stokes he would immediately go to the press with a barrage of unfounded accusations. I refuted all these accusations but he replied “there is nothing you can do, I simply do not trust you”. I was confused why one would turn an incredible source for celebration (of their achievements!) into such bickering, which is when I said that I did not understand why one would risk their career [over unfounded accusations]. Genuinely, at that moment, I was trying to care for him and do a last ditch attempt to get a chance to give them all the credits that they deserve. I deeply apologize for this extremely poor choice of words, it is the opposite of what I was trying to convey. (I should say that I retracted them on the spot by the way.)
Overall, on a personal level, it was incredibly difficult to have these conversations. Levent refused to attend any of the meetings despite my repeated asking. As Sholto Douglas said, there will need to be coordination between Anthropic and OpenAI in the future; I felt I was doing a proxy negotiation with Anthropic while the Anthropic employee refused to directly participate.

See 8 related tweets

@Hadas_Gold: RT @sama: I spent much of the weekend talking with the team who did this work. Seb--and everyone els...
@mikeknoop: > It is true that we tried this because there were rumors on the internet

One general takeaway f...

@teortaxesTex: Levent's gimmick is over. Well, we're all in the same boat, just without Anthropic compensation. Eve...
@teortaxesTex: twitter bullshit made history I don't know who's worse in this story one thing I do know for sure is...
@sama: I spent much of the weekend talking with the team who did this work. Seb--and everyone else--acted w...

10. MTSlive (Group Score: 284.3 | Individual: 49.8)

Cluster: 13 tweets | Engagement: 3771 (Avg: 251) | Type: Tech

SITUATION DETECTED: Mathematicians Tristan Buckmaster and Levent Alpöge have made major progress toward solving the Navier-Stokes existence and smoothness problem, one of the most important problems in mathematics, and say OpenAI may have solved it fully.

See 12 related tweets

@romainhuet: Seeing an OpenAI model make a breakthrough on a problem that has challenged mathematicians for decad...
@mark_k: Rumor: OpenAI may have solved the Navier-Stokes problem, one of the seven Millennium Prize Problems....
@Techmeme: Mathematician Tristan Buckmaster alleges OpenAI learned of his work with Anthropic's Levent Alpöge o...
@cgtwts: openai just killed anthropic’s IPO.\n\nQT @OpenAI: We’re sharing a solution to the Navier-Stokes Mil...
@AravSrinivas: Monumental accomplishment!\n\nQT @OpenAI: We’re sharing a solution to the Navier-Stokes Millennium P...

11. WOLF_Financial (Group Score: 247.7 | Individual: 33.8)

Cluster: 9 tweets | Engagement: 55 (Avg: 62) | Type: Tech

$GOOGL CLOUD AND ACCENTURE LAUNCH 1,000-ENGINEER GEMINI AI DEPLOYMENT GROUP

Google Cloud will train up to 1,000 Accenture forward-deployed engineers to work directly with customers on-site and build AI applications using Gemini Enterprise.

The new group is focused on helping companies redesign business processes, integrate internal data and move AI projects from pilots into production.

Accenture says these teams can take a use case from process mapping to a production-ready AI solution in roughly 8 to 12 weeks.

Source: WSJ

See 8 related tweets

@StockMKTNewz: GOOGLE CLOUD, ACCENTURE LAUNCH UNIT TO PUT AI ENGINEERS ON-SITE WITH CUSTOMERS

Google Cloud $GOOGL...

@MTSlive: SITUATION DETECTED: Accenture and Google Cloud launched the Accenture Gemini Enterprise Business Gro...
@rseroter: Kind of a big deal.

"Building on Accenture’s nearly 50,000 Google Cloud-skilled professionals, the ...

@StockSavvyShay: $GOOGL and$ ACN are building a 1,000 engineer Gemini Enterprise unit to bring agentic AI directly in...
@PolymarketMoney: BREAKING: Google and Accenture are building a “1,000-engineer Gemini Enterprise unit” to bring agent...

12. nebiusai (Group Score: 237.8 | Individual: 40.0)

Cluster: 10 tweets | Engagement: 3729 (Avg: 873) | Type: Tech

We are partnering with Palantir to bring trusted AI cloud infrastructure to Palantir’s commercial customers.

Palantir has named Nebius its preferred sovereign AI infrastructure partner.

Nebius compute and inference endpoints will run inside the Palantir enterprise perimeter – giving customers control over their compute, data, and models.

The partnership is based on a shared vision that open models, continually adapted using customers’ own data, can deliver better domain-specific intelligence while improving control and security.

Read the full press release: https://t.co/OxFxKSOrqr

See 9 related tweets

@wallstengine: $PLTR NAMES$ NBIS ITS PREFERRED SOVEREIGN AI INFRASTRUCTURE PARTNER

Palantir and Nebius are partner...

@amitisinvesting: $PLTR$ NBIS

PALANTIR NAMES NEBIUS AS ITS PREFERRED SOVEREIGN AI INFRASTRUCTURE PARTNER.

Palantir...
@wallstengine: $NBIS +11% https://t.co/QfOXvs8TJ0\n\nQT @wallstengine:$ PLTR NAMES $NBIS ITS PREFERRED SOVEREIGN AI...
@WOLF_Financial: PALANTIR NAMES NEBIUS ITS PREFERRED SOVEREIGN AI INFRASTRUCTURE PARTNER

Palantir $PLTR and Nebius$ ...

@StockSavvyShay: $NBIS is up over 6% after$ PLTR named it its preferred sovereign AI infrastructure partner.

That gi...

13. AdarshChetan (Group Score: 210.7 | Individual: 26.1)

Cluster: 10 tweets | Engagement: 8 (Avg: 51) | Type: Tech

"The goal of AI video shouldn’t necessarily be to remove every constraint.

Some constraints are exactly what make a shot intentional.

Knowing where the camera moves.

Knowing where the character goes.

Knowing how the environment is laid out.

GPT-6 Astra lets those relationships exist in a 3D scene and Viewport before I bring it into CapCut PC for Seedance 2.5 rendering.

Then I can keep refining the footage in the editor.

Maybe creative freedom works better when the important things are already defined.

#GPT6Astra #AIVideo #Seedance25 #CapCutPC"

See 9 related tweets

@NeuraFlowAix: There’s an interesting tradeoff happening in AI filmmaking.

The more AI can improvise, the more imp...

@hey_Jessicaai: RT @WhitmanXAI: Maybe the future of AI filmmaking isn’t one model doing everything.

Think about the...

@AiwithDharmik: RT @Alokkumarzz: One of the most underrated ideas in AI filmmaking might be previsualization.

Tradi...

@AiwithDharmik: RT @supressedvoic: AI video is getting better at generating impressive clips. But I think the more i...
@AnishJaitwar: RT @Grow_withAI: The idea that AI video should be “one click from prompt to finished film” might act...

14. rb_walter_ai (Group Score: 208.7 | Individual: 43.3)

Cluster: 7 tweets | Engagement: 616 (Avg: 117) | Type: Tech

RT @OpenBMB: 🚀 Meet MiniCPM5-2B, a 2B-parameter language model bringing high intelligence density to the edge, now open source!

It ranks #1 among open-source models under 4B parameters on the @ArtificialAnlys Intelligence Index, with a score of 23.

It also scores 20 on the Agentic Index, bringing an early form of general-purpose agent capability to the edge.

Across 34 benchmarks, MiniCPM5-2B achieves an average score of 53.9, covering coding, math, long-context understanding, tool use, and agentic tasks.

And this release goes beyond the model itself.

We’re opening up the data, training recipes, and RL stack behind MiniCPM5-2B.

🤗 Hugging Face: https://t.co/FZOMTZhBjq 💻 GitHub: https://t.co/2L0I8bYv8f Modelscope: https://t.co/WRlPNEAzgu Web: https://t.co/gRSu65FeZL

See 6 related tweets

@JayBisen473370: I don’t think the takeaway from MiniCPM5-2B should be “2B is enough for everything.”

It obviously i...

@ModelScope2022: OpenBMB’s MiniCPM5 brings 128K context, coding, reasoning, and agentic tool use to on-device models....
@OpenBMB: Thank you for the amazing support as always! @ModelScope2022\n\nQT @ModelScope2022: OpenBMB’s MiniC...
@FellMentKE: 2B, open source, strong agentic numbers, and the training stack is open too. This is the kind of rel...
@vinhnx: MiniCPM5-2B, a 2B-parameter language model.\n\nQT @OpenBMB: 🚀 Meet MiniCPM5-2B, a 2B-parameter langu...

15. harvey (Group Score: 199.8 | Individual: 40.7)

Cluster: 7 tweets | Engagement: 246 (Avg: 125) | Type: Tech

We partnered with @baseten to post-train recursive language model (RLM) agents for M&A diligence.

We found that model-harness co-optimization meaningfully improves agent performance in long-horizon environments, bringing us closer to agents that can complete M&A diligence end-to-end.

We started by building an RLM harness that lets a root agent search a data room, delegate document review to sub-agents, and orchestrate the sub-agents' work to produce a final diligence memo.

By distributing document review across sub-agents, the RLM harness can work through thousands of documents in data rooms containing up to 80M tokens.

On LAB Diligence, our synthetic M&A diligence benchmark, the RLM harness raised average rubric criteria pass rate across models from 23% to 62%.

Despite the root agent processing just 1-4% of the tokens in diligence (with sub-agents processing 96-99%), the choice of root agent had a much larger effect on performance than choice of sub-agents.

This led us to focus on the root agent, training two models separately to improve diligence performance.

First, we trained a GLM-5.2 root model in the RLM harness using self-distillation SFT. We selected GLM-5.2 diligence runs with high data room coverage and fine-tuned the model on these trajectories. This increased benchmark performance from 46% to 60%.

Second, we trained a base Qwen3.5-122B-A10B using GRPO within the RLM harness, keeping the sub-agent models fixed. This raised rubric criteria pass rate from 30% to 63% on 50 held-out data rooms.

In both cases we saw major qualitative shifts in agent behavior. The trained root agent learned to delegate more thoroughly and write its memo incrementally as sub-agent findings came back. Data room coverage rose from 62% to 96%. And the root agent learned to scale the size of its agent team proportionally to the size of the data room.

We are now scaling RL training with GLM-5.3, a frontier open-weight model, as the root agent. A small amount of training gives an increase in pass rate from 51% to 59%.

@nikogrupen walks through these experiments with more on harness design, training setup and results, and what we're exploring next:\n\nQT @nikogrupen: https://t.co/Fg8FniKQv3

See 6 related tweets

@nikogrupen: Model-harness co-optimization is helping us solve end-to-end legal use cases like M&A diligence.

To...

@a1zhang: this is so cool!\n\nQT @nikogrupen: Model-harness co-optimization is helping us solve end-to-end leg...
@nikogrupen: RT @gabepereyra: Awesome work by @nikogrupen @ItsJulioPereyra and team with @baseten.

Together with...

@nikogrupen: RT @harvey: We partnered with @baseten to post-train recursive language model (RLM) agents for M&A d...
@baseten: We partnered with @harvey to train agents for M&A diligence and showed that model-harness co-optimiz...

16. JayBisen473370 (Group Score: 195.5 | Individual: 47.1)

Cluster: 7 tweets | Engagement: 437 (Avg: 104) | Type: Tech

INSTEAD OF WATCHING NETFLIX TONIGHT, spend 2 hours learning a skill that could completely change how you work.

This Claude AI FULL COURSE is worth bookmarking.

It teaches you how to use Claude to:

→ Build websites & apps → Write and debug code → Automate repetitive tasks → Analyze files and data → Create AI-powered workflows → Research faster → Build useful projects without starting from scratch → Turn simple ideas into working solutions

And the best part?

You don’t need to be an AI expert.

You just need to understand how to give AI the right instructions and turn it into a tool that actually gets work done.

Most people are using AI like Google:

Ask a question → get an answer → move on.

The people getting the biggest advantage are using it differently:

Idea → AI → workflow → automation → output.

That shift can save hours every single week.

So tonight, you have two choices:

🍿 Spend 2 hours watching Netflix and forget most of it tomorrow.

🧠 Spend 2 hours learning Claude and wake up tomorrow with a skill you can actually use.

If you're a student, developer, creator, marketer, freelancer, founder, or simply someone who wants to stay ahead of AI…

WATCH THIS.

Bookmark it now.

You’ll thank yourself later.

And send it to someone who still thinks AI is just a chatbot.

#ClaudeAI #AI #ArtificialIntelligence #AITools #Automation #Productivity

See 6 related tweets

@AiwithDharmik: INSTEAD OF WATCHING NETFLIX TONIGHT.

Spend 1 hour with this.

Claude AI FULL COURSE that teac...

@JayBisen473370: RT @HarshBisen143: Instead of watching 1 hour of Netflix tonight, watch this Stanford lecture.

It m...

@AiwithDharmik: RT @VikramVerm25510: 🚨 30 AI TOOLS MOST PEOPLE STILL DON’T KNOW ABOUT

Not ChatGPT. Not Claude. Not ...

@rb_walter_ai: RT @Tech_by_Shweta: DON’T WATCH NETFLIX TONIGHT. WATCH THIS INSTEAD. 👀

Give me 2 hours.

This Claud...

@AiwithDharmik: RT @daniel_brook_ai: AI is changing the way we work.

The real advantage isn’t knowing every AI tool...

17. rohanpaul_ai (Group Score: 188.3 | Individual: 23.0)

Cluster: 9 tweets | Engagement: 18 (Avg: 36) | Type: Tech

Cognition raised over $2B at a$ 48B valuation only months after raising over $1B at a$ 26B valuation.

Its reported run-rate revenue rose from $492M to almost$ 900M over the same period, roughly matching the speed of the valuation increase.

This new raise puts Cognition's valuation at roughly 53X its current annualized revenue pace.

thats really an enormous premium on continued growth.

for cotext, among recent AI raises, comparable valuation/revenue multiples are roughly: Wonderful 71x, Lovable 27x, Mistral 24x, Clipto 17x, Socure 14x, Blacksmith 11x, and Higgsfield 8x.

Cognition is therefore at the very high end (53X)\n\nQT @cognition: The world needs far more software than it can build. Cognition exists to change that.

We’ve just raised over $2B at a$ 48B valuation, led by a16z, Accel, Founders Fund, General Catalyst, and Avenir.

Since our round in May, run-rate revenue has grown from $492 M to almost$ 900 M.

See 8 related tweets

@wallstengine: COGNITION RAISES $2B+ AT$ 48B VALUATION AS RUN-RATE REVENUE NEARS $900M

AI coding startup Cognition...

@StockMKTNewz: AI startup Cognition raised more than $2B at a valuation of$ 48 Billion\n\nQT @IPONewsroom_: COGNITI...
@saranormous: Congratulations team @cognition! The independent leader in AI engineering Still early in a new era o...
@IPONewsroom_: COGNITION RAISES $2B+ AT$ 48B VALUATION

The AI software startup behind coding agent Devin has raise...

@eladgil: 🤯\n\nQT @cognition: The world needs far more software than it can build. Cognition exists to change ...

18. Lacoste0x_ (Group Score: 182.8 | Individual: 39.6)

Cluster: 7 tweets | Engagement: 301 (Avg: 88) | Type: Tech

Creators from around the world came together at CapCut World Singapore to share ideas, workflows, and new ways to create.

Next stop: Los Angeles, October 2026 — with AI, creativity, video creation, and the creator community at the center.

#CapCutWorld #CapCutDidThat https://t.co/ytP7SgSYoH

See 6 related tweets

@NovaIAHQ: CapCut World just wrapped in Singapore — bringing creators together to share ideas, workflows, and n...
@JayBisen473370: RT @rb_walter_ai: For creators who are always looking for what’s next. 👀✨

The future of video creat...

@xIrissy: CapCut World is coming to Los Angeles this October 2026! 🌎✨

Bringing creators from around the world...

@OliviaReedai: The future of content creation isn’t just about better tools.

It’s also about bringing creators tog...

@capcutapp: Singapore was just the beginning. See you in Los Angeles!\n\nQT @TheoBuildsAI: The creator world is ...

19. kwindla (Group Score: 169.5 | Individual: 37.6)

Cluster: 5 tweets | Engagement: 37 (Avg: 41) | Type: Tech

One of the lessons from our first couple of years building voice agents is that AI product teams choose a voice entirely based on vibes. This was a bit surprising, given how quantitative so many other agent metrics are.

Voice is emotional. People know the right voice when they hear it.

People often say to us, "I listened to all the voices in the catalog and none of them are quite right."

Gradium's new voice designer makes it easy to iteratively explore the design space, try wildly different things, narrow in on and refine ideas that are promising, and do all of this conversationally by talking to the designer interface, if you want to. (Also, this is really fun. Go try it.)\n\nQT @GradiumAI: Today, we’re launching Voice Design.

Write a prompt, create a voice.

Describe the accent, age, gender, and pace your use case needs, and get new voices in seconds, ready to use.

Live and free in the Gradium API and Studio. https://t.co/wC4YIKwbGQ https://t.co/TlGQWxwAaM

See 4 related tweets

@manthanguptaa: RT @GradiumAI: Today, we’re launching Voice Design.

Write a prompt, create a voice.

Describe t...

@alex_verem: Most voice tools give you safe defaults.

I asked @GradiumAI Voice Design for a Mythical King: deep ...

@alex_prompter: Asked @GradiumAI Voice Design for a battle-worn warrior queen.

Northern English. Dark contralto. Me...

@mattturck: Build and customize any voice you’d like. Live today. 👏👏 @GradiumAI\n\nQT @GradiumAI: Today, we’re...

20. OliviaReedai (Group Score: 165.9 | Individual: 39.8)

Cluster: 6 tweets | Engagement: 142 (Avg: 91) | Type: Tech

Building a $17M ARR business is impressive.

But this is the part GTM teams should pay attention to:

You can now run LinkedIn outreach directly from Claude through HeyReach.

Describe who you want to reach, build the list, draft the sequence and launch it — without doing the outreach manually.

AI integrations get much more interesting when the output is pipeline.

Try HeyReach for free.\n\nQT @nikolak47: Me and Stefan started @heyreach_io from our bedrooms in Macedonia. Buggy af at first.

Since then: $17M ARR and 7,000+ teams running their LinkedIn outreach on it.

This is how you go from “Hey” to deals closed in 2026: https://t.co/BbRzmNJYpS

See 5 related tweets

@shiri_shh: You don’t need to become a LinkedIn creator.

You need clients. Very different assignment.

Keep pos...

@ethancole_ai: RT @OliviaReedai: Building a $17M ARR business is impressive.

But this is the part GTM teams should...

@Parul_Gautam7: $17M ARR from a bedroom is already an impressive journey.

What caught my attention is bringing Link...

@VadimStrizheus: B2B founders after realizing that their entire LinkedIn outbound can now be automated by AI

HeyReac...

@cgtwts: finally, a way to do LinkedIn outreach without actually opening LinkedIn. https://t.co/4Z3dhb75uY\n\...