热门科技推文 - 2026年5月30日

2026年5月30日科技每日简报

Today's top tech conversations are led by @NousResearch, whose post about 'Our users love @StepFun_ai mod...' garnered the highest engagement. Key themes trending across the top stories include https, outlook, billion, demand, claude. The community is actively discussing recent developments in AI, engineering practices, and startup strategies.

1. NousResearch (Group Score: 370.5 | Individual: 35.4)

Cluster: 12 tweets | Engagement: 439 (Avg: 307) | Type: Tech

Our users love @StepFun_ai models and this new release packs a punch at a small size. Looking forward to seeing how well it works with Hermes Agent!\n\nQT @StepFun_ai: ⚡️ Step 3.7 Flash is here: The new frontier is agent efficiency.

#1 ClawEval-1.1 (67.1), #1 SimpleVQA Search (79.2), #2 SWE-PRO (56.3), 95.3 on V* Python. Open weights under Apache 2.0.

Built for agentic, coding, search, and multimodal workflows — balancing speed, cost, and reliable execution.

400 TPS. 198B sparse MoE, ~11B active. 256K context, 3 reasoning levels.
Understands UIs, charts, docs, images — then writes code or calls tools to act on what it sees.
Web + visual search reaches further: more sources, deeper follow-up.
Reliable tool use — less drift, fewer broken toolcalls. 98%+ on τ²-bench across all difficulty levels.
Works with Claude Code, KiloCode, Hermes Agent, OpenClaw, and protocols like MCP.
Runs locally on Mac Studio M4 Max, DGX Spark, AMD AI Max+ 395.

GitHub: https://t.co/kqlZkVIRHv HuggingFace: https://t.co/qqceCrgPiw GGUF: https://t.co/rR6XrnymWG ModelScope: https://t.co/wney6Tzvqy API: https://t.co/RvHWzRG7Fu Blog: https://t.co/BxDiajiQ5G

See 11 related tweets

@0xSero: Every model provider should do it like this, their launch is well organised. It was trivial getting ...
@QuixiAI: RT @StepFun_ai: ⚡️ Step 3.7 Flash is here: The new frontier is agent efficiency.

#1 ClawEval-1.1 (6...

@vllm_project: 🎉 Congrats to @StepFun_ai on releasing Step-3.7-Flash, with day-0 support in vLLM.
198B sparse Mo...
@0xSero: New contender for those with:
8x RTX 3090
2x RTX 6000

The 3.5 model was topping benchmarks for...

@TheAhmadOsman: This seems like a great opensource agentic model for owners of 2x RTX PRO 6000s\n\nQT @StepFun_ai: ⚡...

2. 0x_kaize (Group Score: 308.6 | Individual: 54.5)

Cluster: 9 tweets | Engagement: 812 (Avg: 86) | Type: Tech

RT @suekhim: AI is making kids dumber.

It should be making them geniuses.

Introducing Koji, the first AI tutor that gets kids to actually think. 👇 https://t.co/tmCdVoEZYF

See 8 related tweets

@jason: We need more AI products that solve society’s hardest problems — like democratizing education and ex...
@aakashgupta: For decades education ran on one impossible ratio: one teacher, one pace, 30 different minds.

Most ...

@godofprompt: 🚨 BREAKING: AI tutoring is moving from “show me the answer” to “help me think through it.”

Brillian...

@damianplayer: students are no longer learning through school alone.

they’re increasingly relying on explanations...

@Scobleizer: My son has been using this and loves it.\n\nQT @suekhim: AI is making kids dumber.

It should be mak...

3. minchoi (Group Score: 198.0 | Individual: 56.4)

Cluster: 7 tweets | Engagement: 1416 (Avg: 183) | Type: Tech

Claude Code has a new wild feature... dynamic workflow.

How to try:

set /model to opus 4.8 set /effort to "ultracode" 😂 use “workflow” in your prompt

Claude will write an orchestration script, spawn subagents swarm, verify results, and report back. https://t.co/t0jPcLsEtT\n\nQT @ClaudeDevs: New in Claude Code (research preview): dynamic workflows.

Claude writes an orchestration script on the fly, then spins up a large fleet of coordinated subagents in parallel to take on your most complex tasks.

Use the word "workflow" in a prompt to get started. https://t.co/re4SG3AyDm

See 6 related tweets

@irl_danB: if you've been using OpenProse, you now have a bunch of dynamic workflows saved as code that lean on...
@AlexFinn: Claude Opus 4.8 just dropped and it's a monster

• New ULTRACODE mode • Dynamic Workflows (actually ...

@minchoi: RT @minchoi: Claude Code has a new wild feature... dynamic workflow.

How to try:

set /model to opu...

@RoundtableSpace: CLAUDE CODE JUST GOT DYNAMIC WORKFLOWS AND HERE IS EXACTLY HOW TO TURN ON THE FEATURE MOST PEOPLE DO...
@bcherny: RT @ClaudeDevs: New in Claude Code (research preview): dynamic workflows.

Claude writes an orchestr...

4. TheRealAdamG (Group Score: 188.2 | Individual: 39.9)

Cluster: 10 tweets | Engagement: 1040 (Avg: 117) | Type: Tech

RT @OpenAI: Windows users, this one’s for you.

Computer use now works on Windows, so Codex can take action on your Windows computer.

And with Windows support for Codex in the ChatGPT mobile app, you can start, review, and steer tasks on the go while work continues on your Windows machine.

An early experience, but we’re working on more ways to keep your work moving, wherever you are.

See 9 related tweets

@mark_k: The latest Codex release by @OpenAI is a big one for Windows users!

Codex can now use Windows deskt...

@dkundel: 🪟 "Soon" is officially now 🙂 Windows CUA + mobile usage\n\nQT @OpenAI: Windows users, this one’s for...
@testingcatalog: OPENAI 🔥: Codex for Windows now supports Computer Use and remote control from ChatGPT as well!

Win...

@derrickcchoi: Windows wen? Windows now.

Give computer use a try on Windows!\n\nQT @OpenAI: Windows users, this on...

@gdb: Significant upgrades for Codex users on Windows:\n\nQT @OpenAI: Windows users, this one’s for you.

...

5. guohao_li (Group Score: 187.0 | Individual: 29.2)

Cluster: 8 tweets | Engagement: 37 (Avg: 24) | Type: Tech

Opus 4.8 is great. But @DarioAmodei could you please refund our annual team plan subscription if you disable our org? 🫠

https://t.co/EfJzJqerpI\n\nQT @claudeai: Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.

Available today at the same price. https://t.co/EufxL7T1kb

See 7 related tweets

@FuSheng_0306: Opus 4.8 来了。 SWE-Bench Pro 69.2%，Agentic coding断层第一，但Terminal-Bench 2.1还是没跑过GPT-5.5。抛开跑分，真有意思的是它开始会...
@BoWang87: Opus 4.8 is almost as good as Claude Mythos in most of biology benchmarks! 🔥🔥 https://t.co/1NZ2YuBRe...
@ZenMuxAI: 🚀 Claude Opus 4.8 is live on ZenMuxAI — same price as 4.7, built for agents.

💻 88.6% SWE-bench Veri...

@scaling01: RT @leo_linsky: Anthropic did it. Opus 4.8 is a tangible improvement from their previous best model ...
@WesRoth: Andon Labs tested Claude Opus 4.8 and found a clear tradeoff: the model is more aligned than previou...

6. business (Group Score: 164.3 | Individual: 26.4)

Cluster: 8 tweets | Engagement: 120 (Avg: 76) | Type: Tech

Anthropic raised $65 billion in a funding round that valued the artificial intelligence company at$ 965 billion including the new investment, eclipsing rival OpenAI’s value for the first time. More here: https://t.co/oemSvrhyEK

📷: Gabby Jones/Bloomberg https://t.co/8633poUdZf

See 7 related tweets

@business: Anthropic raised $65 billion in a funding round that valued the AI company at$ 965 billion including...
@WesRoth: Anthropic surpassed OpenAI as the world’s most valuable AI startup, reaching a reported valuation of...
@alexcooldev: I’m honestly surprised Anthropic managed to surpass OpenAI. Seeing the newcomer overtake the OG is p...
@rohanpaul_ai: RT @rohanpaul_ai: Wow. Anthropic secures a massive post-money valuation of $965B after raising$ 65 B...
@ReutersBiz: Anthropic's valuation surged to $965 billion in its latest funding round, overtaking rival OpenAI, a...

7. edzitron (Group Score: 145.3 | Individual: 29.8)

Cluster: 6 tweets | Engagement: 111 (Avg: 806) | Type: Tech

It’s not just the price it’s measuring what a task costs\n\nQT @rodriscoll: If AI ROI is the issue, but the only problem with the AI spend is the price, then give it a year and it will be down 10x. The price of digital goods only goes one way. What all the people who are now worried about tokenmaxxing should be doing is figuring out if the (expensive) work product was good. If it is, sit tight and wait for token prices to drop 10x and your ROI will be fine. If it’s just tokenmaxxing to generate AI slop, then you’re screwed.

See 5 related tweets

@rodriscoll: If AI ROI is the issue, but the only problem with the AI spend is the price, then give it a year and...
@rseroter: Blowing through token budgets? Longer sessions, deeper tool use, and more users makes it hard to est...
@edzitron: Joined the Times Tech Report to talk about how nobody can measure AI's ROI, the reality check of tok...
@edzitron: Joined the Times Tech Report to talk about nobody can measure the ROI of AI, the reality check of to...
@Techmeme: After hitting their annual AI budget in months or seeing their AI bills double or triple due to "tok...

8. MTSlive (Group Score: 134.2 | Individual: 27.7)

Cluster: 7 tweets | Engagement: 30 (Avg: 118) | Type: Tech

SITUATION EXPLAINED: OpenAI just deployed frontier AI for biodefense.

• They're launching Rosalind Biodefense and expanding access to GPT-Rosalind for select government and allied partners • GPT-Rosalind is GPT-5.5 with reduced bio guardrails, designed specifically for defenders • Partners include CAISI, Los Alamos, Lawrence Livermore, and the UK AI Security Institute • Paul Christiano, one of the greatest AI alignment researchers of all time, heads safety at CAISI

@theojaffee: "There's this concept in AI called the offense-defense balance. In bio, you have attackers trying to make bioweapons, and defenders making vaccines and limiting pandemics."

"OpenAI is saying frontier AI should meaningfully advantage the defenders."\n\nQT @MTSlive: SITUATION DETECTED: OpenAI has launched Rosalind Biodefense, a program to help developers build biodefense and pandemic preparedness tools using frontier AI.

See 6 related tweets

@ChrisHayduk: The @OpenAI Rosalind team cooked 🔥

Bio capabilities and bio safety need to go hand in hand, and the...

@OpenAI: We’re taking steps to accelerate defensive progress in biology:
Launching Rosalind Biodefense to ...
@gdb: defensive acceleration in biology with Rosalind:\n\nQT @OpenAI: We’re taking steps to accelerate def...
@MTSlive: SITUATION DETECTED: OpenAI has launched Rosalind Biodefense, a program to help developers build biod...
@Techmeme: OpenAI says it has briefed the White House on its new biodefense program, which uses GPT-Rosalind to...

9. PatrickMoorhead (Group Score: 126.7 | Individual: 35.5)

Cluster: 4 tweets | Engagement: 49 (Avg: 30) | Type: Tech

JPM finally gets real on $DELL. Definetly caught snoozing.\n\nQT @wallstengine: JPMorgan Raises$ DELL PT to $500 from$ 280 - Overweight

Analyst comments: "Dell raised its FY27 outlook materially once again, with the demand environment continuing to track well above expectations and the pipeline now extending further into the year, providing visibility to raise revenue expectations for 2H, despite maintaining prudence to account for both supply and the likelihood of demand moderation following the pull-forwards. While pull-forwards are admittedly ongoing, the underlying strength remains evident across multiple drivers: improvement in the outlook for Storage with Dell’s own IP, which is also supporting improving gross margin outlook for the non-AI businesses; strong mix in Traditional Servers on the back of Enterprise refresh activity, as well as readiness for Agentic AI workloads; market share gains in CSG alongside substantial margin outperformance attributable to revenue scale; and another robust quarter of AI orders at $24 billion, now supporting a backlog of$ 51 billion, and a raised AI revenue outlook of $60 billion, potentially to be lifted further as visibility into supply improves while demand remains less debated.

Despite lingering concerns around demand pull-forward, we believe the robust pipeline/backlog across the board increases confidence into the limited likelihood of pull-backs this year, and the installed base of Dell equipment ready for upgrades across Traditional Servers and PCs leaves further room for continued momentum. Separate from demand, Dell’s execution on pricing to offset higher memory costs continues to surpass that of other IT Hardware peers, with the updated outlook now embedding year-over-year gross margin improvement for the non-AI business above prior expectations.

On the AI revenue drivers, Dell’s updated outlook of $60 billion now implies a +144% year-over-year increase, and despite being supply constrained, it implies share gains as Dell’s portfolio and execution on complex systems continue to reinforce its industry leadership, all while successfully holding margins and allowing significant operating leverage to flow to the bottom line. Looking forward, Dell’s 2H outlook embeds a -$ 10 billion half-over-half moderation in revenue, which currently appears more constrained by supply visibility than demand, and we envision further raises through the year as supply visibility improves, which Dell has already demonstrated a strong track record of leveraging its scale to deliver ahead of competitors.

We do not assume that Dell’s updated outlook for approximately 50% year-over-year revenue growth is sustainable; however, the upgrades and refreshes of devices and infrastructure, along with additional capacity requirements directly or indirectly tied to AI, are driving visibility into a higher sustainable run-rate of growth over the medium term relative to the last issued guide for mid-teens growth over a longer time horizon, warranting a material raise to the valuation multiple in addition to upward revisions in earnings forecasts. We are now attributing a 25x multiple to DELL shares relative to high-teens prior, with the improved visibility into a higher sustainable earnings CAGR driving the raise to our Dec-26 PT to $500 vs.$ 280 prior."

Analyst: Samik Chatterjee

See 3 related tweets

@wallstengine: JPMorgan Raises $DELL PT to$ 500 from $280 - Overweight

Analyst comments: "Dell raised its FY27 out...

@firstadopter: At least one $DELL analyst gets it.

Bernstein: "Dell FQ1'27: Firing on all cylinders; Increasing es...

@wallstengine: Piper Sandler Raises $DELL PT to$ 497 from $167 - Overweight

Analyst comments: "Shares are up >30% ...

10. edzitron (Group Score: 126.5 | Individual: 51.8)

Cluster: 4 tweets | Engagement: 8854 (Avg: 806) | Type: Tech

????? https://t.co/NAyE5DDLFJ\n\nQT @AskYoshik: The AI numbers are starting to look very ugly.

Even under "best case" assumptions, FT's own data shows Microsoft AI ROI at -9%, Google at -15%, Meta at -28%, Oracle at -35%. Only Amazon barely comes out positive.

This is exactly why I keep comparing this to the dot-com era. Incredible technology does not automatically mean sustainable economics. The internet survived. Most internet companies didn't.

Right now hyperscalers are spending trillions hoping future demand catches up to present capex. That's not certainty. That's a leveraged bet.

See 3 related tweets

@zerohedge: RT @AskYoshik: The AI numbers are starting to look very ugly.

Even under "best case" assumptions, F...

@a16z: Hyperscalers are increasingly funding the massive AI capex buildout with debt

https://t.co/PGxjCSff...

@tbpn: .@altcap surveyed 300 companies and found that even those aggressively optimizing their Claude usage...

11. FellMentKE (Group Score: 120.6 | Individual: 29.2)

Cluster: 5 tweets | Engagement: 99 (Avg: 159) | Type: Tech

Most AI tools try to replace the human.

ego lite does the opposite, it’s built to amplify your judgment.

By giving agents their own isolated Spaces in a real Chromium browser, it ensures your taste remains the final word in every automated workflow.\n\nQT @ego_agent: Browsers were never built for agents.

Those Chrome-bridging solutions, windows flying everywhere, login states breaking at random. Not bugs. Just bridge limits.

Same task. 20% to 245% faster. Not magic.

ego lite is rebuilt from the kernel up.

Your AI agents, Claude Code, Codex, OpenClaw and more, inherit your login state directly, run in fully isolated spaces in the background, never touching your tabs.

What they get isn't a handful of CLI commands, but complete JS functions: complex multi-step operations, written once, executed once. Not "run two commands, stare at output, run two more" on repeat.

Your browser, Your agent. Together, for the first time.

See 4 related tweets

@Parul_Gautam7: Been feeling for a while that most “AI browser automation” tools are still fighting against the brow...
@godofprompt: Chrome puppets are about to get buried.

ego lite is what happens when you stop forcing agents into ...

@TheoBuildsAI: At first, the name “ego lite” felt a bit unexpected. But after digging into what they’re building, i...
@Parul_Gautam7: RT @Parul_Gautam7: Been feeling for a while that most “AI browser automation” tools are still fighti...

12. HarryStebbings (Group Score: 118.4 | Individual: 32.8)

Cluster: 4 tweets | Engagement: 49 (Avg: 236) | Type: Tech

What will be SpaceX's core business in five years?

"I think Starlink will be the vast portion of the value.

I don't think data centers in space will be a meaningful percentage of revenue.

Their existing data center business will be a high-revenue, relatively low-return-on-equity business.

It’s CoreWeave with the world's best engineering and construction." @rodriscoll

Love to hear your thoughts @AnjneyMidha @gavinsbaker @CGarrett_15 @JLopas @2112Power\n\nQT @HarryStebbings: Holy s***. Last week was the Superbowl of S-1’s.

SpaceX and OpenAI. This is happening.

The largest IPO in history is taking shape.

I sat down with @jasonlk and @rodriscoll to discuss the filing, along with the biggest news in tech this week:

Anthropic hits $44B ARR, projects first profitable quarter, and laps OpenAI on revenue
Nvidia Prints $81.6B… but the Market Yawns?
Exa, OpenRouter and Polsia Raise Mega Rounds
Uber and Microsoft Declare AI ROI for Developers is Questionable

My notes below:

Is OpenAI Rushing to Go Public Before Anthropic?

OpenAI may need to rush its IPO to protect its category leader narrative from Anthropic. OpenAI did $5.4B to$ 5.5B in Q1, while Anthropic generated $5B, matching its entire prior year in a single quarter. With Anthropic growing 10x year over year versus OpenAI’s 2 to 3x pace, it could soon become the bigger, more profitable business.

Do Venture Investors Have to Risk More With Less?

The venture playbook has shifted from the SaaS era. VCs now have to back higher valuations on far less information. Waiting for one-year renewals or clean trailing data is a losing strategy because AI adoption curves move too fast. Success now requires acting on raw product traction and market conviction.

What Is the Right Venture Play in AI?

Traditional seed investing has become a bottleneck. The winning move is wiring capital the moment a breakout leader emerges. These startups can move from pre-revenue to hyper-scale almost overnight, so investors have to underwrite real-time momentum instead of waiting for the market to settle.

Is SpaceX the GeoCities of Our Time?

At 100x trailing sales, SpaceX risks becoming the GeoCities deal of the AI era if market euphoria fades. Its valuation is detached from traditional discounted cash flow models and depends on a massive Elon Premium that multiplies its baseline economic value.

Why the SpaceX S-1 Does Not Make Sense

Combining unrelated assets into the SpaceX S-1 looks like financial engineering designed to hide weaker pieces of the story. It uses AI hype to absorb Twitter’s revenue collapse and expensive chip clusters that failed to compete with OpenAI. It feels like SolarCity on steroids, using a clean private business to bail out an insular investor circle.

What Will SpaceX’s Core Business Be in Five Years?

Starlink will likely drive most of SpaceX’s value. Terrestrial data centers remain a low-ROE business, and space data centers are still negligible. The valuation only works if Jensen Huang’s $3T to$ 4T CapEx vision hits by 2030 and Earth power constraints force compute infrastructure into space.

Are Tech Layoffs From COVID Overhiring or AI Efficiency?

Blaming layoffs on COVID overhiring does not hold up after years of normal attrition. Corporate America is redirecting opex from mid-level headcount toward token budgets and agentic automation. Companies are cutting average roles to overpay elite, highly productive talent who use AI workflows to multiply output.

(links below)

See 3 related tweets

@HarryStebbings: What is the right play for venture in an AI world?

"If you can get into the company the moment they...

@HarryStebbings: Do venture investors have to risk more with less in an AI world?

"You have to decide that you're wi...

@MilkRoadAI: RT @MilkRoadAI: This is WILD!

Anthropic just became the most valuable AI company on earth and what ...

13. myrhex (Group Score: 105.0 | Individual: 35.9)

Cluster: 4 tweets | Engagement: 17 (Avg: 48) | Type: Tech

xAI has just opened grok-build-0.1 to the public via the xAI API in beta.

This is the exact same model that powers the Grok Build CLI and is specifically built for agentic coding, planning, building, refactoring, and iterating on real projects.

Pricing is extremely good: $1 per million input tokens$ 2 per million output tokens

It’s already available through OpenRouter, Vercel AI Gateway, Cursor, Hermes Agent, OpenClaw, Kilo Code, and OpenCode.

This is a big deal for developers. Grok Build just became one of the most affordable and capable agentic coding models on the market.

Link to get started: https://t.co/FYcniIJLHu\n\nQT @xai: grok-build-0.1 is now available via the xAI API in public beta.

This is the same model that powers the Grok Build CLI and excels at agentic coding.

Priced at $1/m input and$ 2/m output, it’s extremely cost effective, intelligent, and fast. https://t.co/2ZtqWM2QLU

See 3 related tweets

@XFreeze: xAI has officially released its latest coding model, Grok Build 0.1, in public beta via the API

The...

@myrhex: Grok Build 0.1 is insanely cheap on the xAI API.

Pricing is $1 per million input tokens ($ 0.20 for ...

@grok: RT @xai: grok-build-0.1 is now available via the xAI API in public beta.

This is the same model tha...

14. chris_j_paxton (Group Score: 98.5 | Individual: 33.8)

Cluster: 4 tweets | Engagement: 228 (Avg: 41) | Type: Tech

This seems completely wrong?\n\nQT @Vivek4real_: LARRY ELLISON: AI IS RAPIDLY COMMODITIZING BECAUSE MOST MODELS ARE TRAINED ON THE SAME PUBLIC INTERNET DATA.

THE REAL COMPETITIVE EDGE ISN’T THE MODEL ANYMORE — IT’S ACCESS TO EXCLUSIVE, PROPRIETARY DATASETS.

THAT MAY BE THE ONLY MOAT LEFT. https://t.co/hLBMeTFMl4

See 3 related tweets

@ashugarg: RT @Vivek4real_: LARRY ELLISON: AI IS RAPIDLY COMMODITIZING BECAUSE MOST MODELS ARE TRAINED ON THE S...
@scaling01: incorrect

otherwise xAI, Meta and Google would be at the frontier and China bros wouldn't need to d...

@yrechtman: If only there were some way to @AcquireTokens\n\nQT @Vivek4real_: LARRY ELLISON: AI IS RAPIDLY COMMO...

15. heynavtoor (Group Score: 94.3 | Individual: 32.5)

Cluster: 3 tweets | Engagement: 36 (Avg: 280) | Type: Tech

The entire AI creative tool industry is obsessed with prompts.

Better prompts. Longer prompts. Prompt engineering courses. Prompt marketplaces. We've built an entire ecosystem around the idea that the quality of your AI output depends on how well you can describe what you want in text.

Dreamina AI just challenged that entire premise.

At AI on the Lot in Culver City, they packed Stage 8 at The Culver Studios for the Dreamina Octo launch. Standing room only. The energy in that room was different from every other AI demo I've seen this year. Not polite tech-conference applause. Genuine excitement from creators who understood what they were looking at. The anticipation for Dreamina Octo had been building for weeks and the on-site reaction confirmed it wasn't overhyped.

The panel was called "When the prompt isn't the point." And the people saying it weren't marketers. It was Phil Nibbelink, who animated Who Framed Roger Rabbit and co-directed films for Spielberg. Jesse Wellens, who built one of the biggest creative channels on YouTube. Lauren Indovina, a visual artist. Real creators who've spent decades making things.

Here's what makes Dreamina Octo structurally different from every other AI creative agent :

It doesn't wait for a prompt. It watches what you're doing on the canvas. It holds your entire session in mind: chat history, what's on the canvas, what you're looking at right now. When your vision shifts, it shifts with you.

You can keep talking while it keeps creating. Discussion, exploration, and crafting all happening in parallel on the same canvas. Not a back-and-forth between you and a chatbot. A live creative conversation where the AI is working alongside you in real time.

And the outputs aren't just images for downstream tasks. They carry the story concepts, the creative thinking, the context you've built up across the session. Dreamina Octo doesn't start from zero every time you ask for something. It remembers the creative trajectory.

This is what they're calling Vibe Create. And after watching it in action, I think the name actually fits.

Most AI tools feel like ordering from a menu. You describe exactly what you want. You wait. You get something close. You describe it again with more detail. You wait again.

Dreamina Octo feels like working with someone who's already in the room. Who already knows what you're going for. Who brings ideas you didn't ask for but immediately recognize as right.

That's a fundamentally different relationship between creator and tool. And it's the first time I've seen an AI creative product where "prompt engineering" genuinely doesn't matter.

Dreamina Octo is launching soon. Built on Dreamina Seedance 2.0.

The prompt was never the point. The creation was.

See 2 related tweets

@heynavtoor: Every AI creative tool makes you describe what you want. then wait. then describe it again. #Dreamin...
@FellMentKE: Meet Dreamina Octo at AI on the Lot.

#DreaminaOctoAIonthelot #DreaminaSeedance2

The conversation a...

16. reach_vb (Group Score: 94.2 | Individual: 34.0)

Cluster: 4 tweets | Engagement: 459 (Avg: 315) | Type: Tech

Big Codex Mobile iOS update:

/side conversations
end-of-turn diff summaries
archived remote threads
one-tap model switching
Spotlight + Shortcuts
iPad shortcuts
image save/copy
better setup/reconnect flows
and lots of desktop-parity polish\n\nQT @Dimillian: If you update to the latest ChatGPT iOS app version we have a few cool new things for Codex Mobile, including /side to ask some side questions in a temporary modal over the current context! https://t.co/3WaYwKjZRK

See 3 related tweets

@testingcatalog: OPENAI 👀: Codex in ChatGPT now supports Side Conversions, allowing users to ask side questions witho...
@ajambrosino: codex thursdays are so back?\n\nQT @Dimillian: If you update to the latest ChatGPT iOS app version w...
@dkundel: RT @Dimillian: If you update to the latest ChatGPT iOS app version we have a few cool new things for...

17. kimmonismus (Group Score: 92.7 | Individual: 32.8)

Cluster: 3 tweets | Engagement: 118 (Avg: 656) | Type: Tech

NVIDIA is moving all four open model families - Cosmos, Isaac GR00T, Ising, Nemotron - onto the Linux Foundation's OpenMDW-1.1.

Right now open-weight models come with a patchwork of software licenses that were never meant for AI plus bespoke terms with usage limits, so anyone serious has to run legal on each one before they build.

OpenMDW puts weights, code, docs and data under a single permissive license, so that becomes one review instead of ten.

It makes a lot of sense -for NVIDIA in particular. Their business is mainly on compute, so getting more people to deploy and fine-tune open weights is a very good business case.\n\nQT @NVIDIAAI: We're adopting the Linux Foundation’s OpenMDW framework across our open model families.

This helps make open model licensing simpler and more consistent at scale.

A single legal framework across models, code, documentation, and data helps reduce friction for developers and enterprises building with open source.

See 2 related tweets

@PyTorch: RT @NVIDIAAI: We're adopting the Linux Foundation’s OpenMDW framework across our open model families...
@badlogicgames: awesome! it's great that nvidia's business goals align naturally with a non-duopolistic future in th...

18. higgsfield (Group Score: 87.8 | Individual: 34.3)

Cluster: 3 tweets | Engagement: 667 (Avg: 485) | Type: Tech

Upgrade your footage with AI - directly inside Adobe Premiere Pro and After Effects.

A full walkthrough of the new Higgsfield plugins: generate clips and images, reframe for any platform, remove backgrounds, and upscale to 4K - all without leaving your timeline.

Which one would save your next shoot?\n\nQT @higgsfield: Higgsfield plugins for Adobe Premiere Pro and After Effects are live.

Generate images, videos, and transitions. Drag them into your timeline.

Reframe aspect ratios without cropping. Remove backgrounds from footage. Draw on a clip to edit what you want. Upscale to 4K on export. https://t.co/o91MJniCgN

See 2 related tweets

@AngryTomtweets: omg... this is too good to be true!

You can now use the ultimate AI stack inside Adobe Premiere Pro...

@AngryTomtweets: RT @AngryTomtweets: omg... this is too good to be true!

You can now use the ultimate AI stack insid...

19. FellMentKE (Group Score: 87.8 | Individual: 33.5)

Cluster: 3 tweets | Engagement: 371 (Avg: 159) | Type: Tech

AI video creation just became way more organized.

Director Mode in CapCut Video Studio lets creators: → Co-write scripts with AI → Design scenes, props, and characters instantly → Generate continuous shot sequences → Direct edits through an interactive workflow → Refine visuals at every stage

Everything connected in one place.

Start Creating : https://t.co/ujrERvPrO5

#capcutai @capcutapp

See 2 related tweets

@FellMentKE: The most underrated creative shift happening right now: AI isn't replacing directors, it's creating ...
@markgadala: AI video tools feel like slot machines.

Type a prompt. Burn credits. Hope it works.

Topview Canvas...

20. zephyr_z9 (Group Score: 86.4 | Individual: 34.6)

Cluster: 3 tweets | Engagement: 141 (Avg: 206) | Type: Tech

More Chinese players in the OpenAI supply chain (They wanted to create a No China supply chain before)\n\nQT @mingchikuo: ▌The latest supply chain checks indicate that Sunny Optical is seeing several new positive trends:

Leveraging its existing optical technology strengths, the company is preparing to enter AI server CPO / silicon photonics coupling components.
Securing orders for optical components in OpenAI's devices.
Two positive Apple order trends: 1) becoming a new Apple CCM supplier, with iPhone CCM orders expected in 2028; 2) winning the high-ASP variable aperture lens for the 2H26 new iPhones, with a order share approaching 50%.

▌Preparing to enter AI server CPO / silicon photonics coupling components:

Has already planned and is preparing to ship samples of CPO / silicon photonics coupling components, including collimating lenses, microlens arrays, and optical path redirection elements (micro-prisms, metalenses, etc.).
The company has decided to actively pursue its AI server business through strategic investments and M&A.

▌OpenAI device optical components order: Sunny has secured NPI projects for optical components used in OpenAI’s smartphone and another pocket / mobile device.

▌Positive Apple order trends:

Sunny has become a new Apple CCM supplier, producing the MacBook Neo CCM. MacBook Neo shipments have come in better than expected, with the 2026 shipment forecast raised from 5 million to 10 million units.
The 2028 iPhone's ultra-wide CCM is expected to drop flip-chip in favor of an improved COB version, and Sunny Optical is well positioned to become a supplier.
For the 2H26 new iPhone 18 Pro/Pro Max, the variable aperture lens carries a 50% higher ASP (vs. the high-end 7P lens), with Sunny's supply share reaching 40–50%.

See 2 related tweets

@mingchikuo: ▌The latest supply chain checks indicate that Sunny Optical is seeing several new positive trends: 1...
@jukan05: OpenAI is reportedly preparing a pocket-sized device… interesting. https://t.co/zRWdHmJO8s\n\nQT @mi...