科技推特精选 - 2026年3月6日

2026年3月6日科技每日简报

Today's top tech conversations are led by @rseroter, whose post about 'RT @addyosmani: Introducing th...' garnered the highest engagement. Key themes trending across the top stories include https, companies, shipper, google, openai. The community is actively discussing recent developments in AI, engineering practices, and startup strategies.

1. rseroter (Group Score: 183.6 | Individual: 56.1)

Cluster: 5 tweets | Engagement: 2308 (Avg: 78) | Type: Tech

RT @addyosmani: Introducing the Google Workspace CLI: https://t.co/8yWtbxiVPp - built for humans and agents.

Google Drive, Gmail, Calendar…

See 4 related tweets

@burkeholland: RT @rauchg: Google has shipped a CLI for Google Workspace (Drive, Gmail, Calendar, Sheets, Docs, …) ...
@addyosmani: Introducing the Google Workspace CLI: https://t.co/8yWtbxiVPp - built for humans and agents.

Google...

@dair_ai: RT @omarsar0: Banger CLI tool released by Google.

CLI for Google Workspace + a bunch of useful Agen...

@nummanali: The new Google Workspace CLI was first posted by @kazunori_279

If you posted, at least give him cr...

2. martin_casado (Group Score: 129.1 | Individual: 41.8)

Cluster: 9 tweets | Engagement: 4740 (Avg: 440) | Type: Tech

RT @OpenAI: GPT-5.4 Thinking and GPT-5.4 Pro are rolling out now in ChatGPT.

GPT-5.4 is also now available in the API and Codex.

GPT-5.4…

See 8 related tweets

@sherwinwu: GPT-5.4 Thinking is out!

If you thought GPT-5.3-Codex was good, just wait until you give 5.4 a spi...

@cryps1s: RT @sama: GPT-5.4 is launching, available now in the API and Codex and rolling out over the course o...
@RoundtableSpace: JUST IN: GPT-5.4 THINKING AND GPT-5.4 PRO ARE ROLLING ARE OUT NOW IN CHATGPT. https://t.co/XlKCvUIFM...
@TechCrunch: OpenAI launches GPT-5.4 with Pro and Thinking versions https://t.co/Q2GJPk2hgH...
@testingcatalog: BREAKING 🚨: OPENAI IS ROLLING OUT GPT-5.4 ACROSS CHATGPT AND CODEX APPS! https://t.co/R7bohvVhPz...

3. chatgpt21 (Group Score: 113.8 | Individual: 31.7)

Cluster: 4 tweets | Engagement: 101 (Avg: 140) | Type: Tech

“Compared with the previous GPT-5.2 model, GPT-5.4 shows improvements in:

•Coding, document understanding, tool use, and instruction following

•Image perception and multimodal tasks

•Long-running task execution and multi-step agent workflows

•Token efficiency and end-to-end performance on tool-heavy workloads

•Agentic web search and multi-source synthesis, especially for hard-to-locate information

•Document-heavy and spreadsheet-heavy business workflows in customer service, analytics, and finance”

See 3 related tweets

@reach_vb: BOOOOM! Introducing GPT-5.4 Thinking & Pro in Codex, API & ChatGPT 🔥

It combines GPT-5.3-Codex-leve...

@mark_k: Today, the team at @OpenAI released GPT-5.4, a model designed to improve efficiency in professional ...
@OpenAIDevs: See what developers are building with GPT-5.4.

Our developer app showcase highlights examples acros...

4. rohanpaul_ai (Group Score: 109.3 | Individual: 19.2)

Cluster: 7 tweets | Engagement: 132 (Avg: 122) | Type: Tech

🚨 BREAKING: The Pentagon and Anthropic are back at the table discussing AI again, according to FT.

This last-ditch effort aims to prevent Anthropic from being officially labeled a "supply chain risk" and banned from government work.

Dario Amodei, the CEO of Anthropic, is meeting with top defense officials to compromise on how the military can use their AI models.

ft .com/content/97bda2ef-fc06-40b3-a867-f61a711b148b

See 6 related tweets

@business: RT @KatrinaManson: SCOOP: The Pentagon has formally notified Anthropic that it’s deemed the artifici...
@techdaily24: SHOCKER: Pentagon labels Anthropic (Claude AI creator) a supply chain risk, banning federal business...
@business: The Pentagon says it has formally notified Anthropic that it’s deemed the AI company and its product...
@Reuters: 🔊 'This particular designation essentially says no Pentagon contractor can use Anthropic's AI in the...
@Cointelegraph: 🇺🇸 JUST IN: Anthropic CEO Dario Amodei is back in talks with the Pentagon to reach a compromise on m...

5. Reuters (Group Score: 91.2 | Individual: 44.9)

Cluster: 4 tweets | Engagement: 2508 (Avg: 289) | Type: Tech

Google, Microsoft, Meta, Amazon and several artificial intelligence companies signed a pledge at the White House to bear the cost of new electricity generation to power their data centers https://t.co/cZDfq9de8i https://t.co/M2wssk1LIy

See 3 related tweets

@ReutersBiz: WATCH: Google, Microsoft, Meta, Amazon and several artificial intelligence companies signed a pledge...
@Cointelegraph: 🇺🇸 HUGE: Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI have signed the Ratepayer Protecti...
@arstechnica: Leading AI, datacenter companies sign pledge to buy their own power https://t.co/tPSzd8nLAZ...

6. chatgpt21 (Group Score: 84.9 | Individual: 32.3)

Cluster: 5 tweets | Engagement: 107 (Avg: 140) | Type: Tech

When I started hearing employee chatter about GPT 5.4 I thought they were joking.

But it is undeniable OpenAI is shipping more models faster than anyone else right now, (while hopefully scaling pre training in the background)

Crazy to think we’re not even 2 whole number iterations past 3.5

The next update presumably after 5.4 codex, will be GPT 5.5 and they have to make a big mark as it’s a .5 iteration

We will officially be 2 whole digit iterations past GPT 3.5 and it’s insane to think how much progress we’ve made since 3.5 and we’re only on 5.5..

Can’t imagine how good GPT 7.5 or 8.5 will be

See 4 related tweets

@scaling01: For some reason I dreamt that OpenAI is going to release two open-source models today

but I guess G...

@scaling01: OpenAI said they are moving towards monthly model releases

this implies:

GPT-5.5 in April
GPT...
@dejavucoder: openai model launches not as exciting as claude releases for some reason. nevertheless excited to tr...
@scaling01: RT @petergostev: BullshitBench v2 update: GPT-5.4 is ahead of other OpenAI models (slightly), but is...

7. mark_k (Group Score: 80.1 | Individual: 32.5)

Cluster: 3 tweets | Engagement: 328 (Avg: 220) | Type: Tech

OpenAI is developing a new bidirectional audio model designed to make voice assistants feel significantly more natural. According to a report from The Information, the update aims to eliminate the robotic, turn-based style of current AI conversations.

While @ChatGPTapp's current voice mode often stops abruptly when interrupted, this upcoming model continuously processes audio in the background. It can register interjections and adjust its response on the fly, much like a human would.

This fluidity could be a major breakthrough for complex interactions. For example, a customer service agent could seamlessly handle a caller changing their mind mid-sentence without losing the flow of the conversation.

We may have to wait a bit longer to use it. The prototype reportedly still faces glitches during longer sessions, suggesting the release has likely been pushed back to the second quarter of the year or later.

Source: @theinformation

See 2 related tweets

@kimmonismus: OpenAI Developing Real-Time Conversational Audio

OpenAI is building a new bidirectional (BiDi) audi...

@testingcatalog: OpenAI is working on a bidirectional voice model that can adjust its response in realtime, according...

8. business (Group Score: 78.8 | Individual: 18.5)

Cluster: 5 tweets | Engagement: 104 (Avg: 147) | Type: Tech

TUNE IN: How can investors, businesses and workers navigate the "AI scare trade" that is spooking markets? Bloomberg's @timsteno speaks with @EricJWeiner1 @rachelmetz and Ellen DiMauro — answer your questions in real time.

Join the conversation: https://t.co/wetBjdeUDw https://t.co/R8oJ4MZ2a5

See 4 related tweets

@business: HAPPENING SOON: How can investors, businesses and workers navigate the "AI scare trade" that's swing...
@business: JOIN THE CONVERSATION: How can investors, businesses and workers navigate the "AI scare trade" that ...
@business: 🎧 LIVE NOW: How can investors, businesses and workers navigate the "AI scare trade" that's swinging ...
@business: TODAY: How can investors, businesses and workers navigate the "AI scare trade" that is spooking mark...

9. chddaniel (Group Score: 71.9 | Individual: 51.9)

Cluster: 2 tweets | Engagement: 288 (Avg: 37) | Type: Tech

BREAKING: @claudeai just got a huge upgrade today and I'm very happy to put it together in shipper. As of now, Claude Opus 4.6 can create self-building companies for you.

We just launched Shipper 2.0 a power-up that lets Claude:

✅ Build web/mobile apps and Chrome extensions ✅ Code, design, monetize, launch ✅ Do email marketing ✅ Self-maintain the business

Opus 4.6 can now do all the above from a <10 word prompt and for as low as $0.12/app... What used to take hours and hundreds of thousands in salaries now simply takes 1 prompt.

Go in Shipper and ask Claude to "build a talent hiring platform" or "a calorie tracking app like MyFitnessPal".

To celebrate, we're giving away free credits randomly to people who repost and comment "SHIPPER' :)

See 1 related tweets

@RoundtableSpace: Shipper 2.0 lets Claude Code build your app, design it, monetize it, do your email marketing, and se...

10. rohanpaul_ai (Group Score: 63.3 | Individual: 48.3)

Cluster: 2 tweets | Engagement: 1284 (Avg: 122) | Type: Tech

Mark Cuban on the next job wave.

Customized AI integration for small to mid-sized companies.

"Software is dead because everything's gonna be customized to your unique utilization. Who's gonna do it for them... And there are 33 mn companies in the US."

https://t.co/JczlPMOC1C

See 1 related tweets

@rohanpaul_ai: RT @rohanpaul_ai: Mark Cuban on the next job wave.

Customized AI integration for small to mid-sized...

11. omarsar0 (Group Score: 63.3 | Individual: 24.7)

Cluster: 4 tweets | Engagement: 77 (Avg: 118) | Type: Tech

Just switched one of my proactive agents to Codex.

Absolutely insane setup.

Great for high level and complex orchestration.

GPT-5.4 is just going level things up fast. Makes sense with those advanced reasoning capabilities.

See 3 related tweets

@OpenAIDevs: Codex got more speed.

With /fast mode, GPT-5.4 runs 1.5x faster with the same intelligence and reas...

@romainhuet: Hello GPT-5.4! Massive improvements in computer-use and agentic workflows.

Great news for Codex bui...

@TheRealAdamG: RT @Rasmic: The way gpt 5.3 codex is cooking for is insane... it just gets me. It just understands....

12. JamesMontemagno (Group Score: 60.9 | Individual: 22.0)

Cluster: 3 tweets | Engagement: 123 (Avg: 2306) | Type: Tech

This new release of @code is INCREDIBLE.

One of my favorite new features is the new integrated browser with agent mode integration. It is 🔥. https://t.co/1o44eoS7nh

See 2 related tweets

@JamesMontemagno: This new release of @code is INCREDIBLE.

One of my favorite new features is the new integrated brow...

@pierceboggan: RT @JamesMontemagno: This new release of @code is INCREDIBLE.

One of my favorite new features is th...

13. cryptopunk7213 (Group Score: 55.8 | Individual: 25.3)

Cluster: 3 tweets | Engagement: 116 (Avg: 915) | Type: Tech

GPT 5.4 is officially better than most people at doing their jobs!

its a BEAST at excel / spreadsheet work, bye bye 50% of finance work.
its really fckin good at using a computer, officially beats humans in tests.
coding: anthropic's claude, still holds the crown here, but not by much. gpt 5.4 can throw down.
33% less hallucination = doesnt bullshit you
#1 in scientific reasoning
1M context window (slam your entire codebase/book into it) but 47% FEWER tokens = its cheap af
you can interrupt its thinking to steer it your own way.

overall solid model update imo, nothing groundbreaking but i am so fucking psyched at how quickly these models are coming out

See 2 related tweets

@QuixiAI: GPT-5.4 is really good.

I immediately notice its boost in understanding and ability to solve proble...

@sdrzn: gpt 5.4 is better at using computers than humans now...

we are living in sci-fi times...

14. Reuters (Group Score: 53.5 | Individual: 19.5)

Cluster: 3 tweets | Engagement: 27 (Avg: 289) | Type: Tech

From Breakingviews - Breakingviews - War angst and AI hype mar Seoul's market ambitions https://t.co/j2XkWfqjPC https://t.co/j2XkWfqjPC

See 2 related tweets

@Reuters: Breakingviews - War angst and AI hype mar Seoul's market ambitions https://t.co/ctRCtihdOA https://t...
@Breakingviews: War angst and AI hype mar Seoul's market ambitions
https://t.co/QhbXcFqBXd https://t.co/Z2bABNzEg...

15. rohanpaul_ai (Group Score: 53.4 | Individual: 36.0)

Cluster: 2 tweets | Engagement: 146 (Avg: 122) | Type: Tech

Bloomberg: Investors on Wall Street have a new fear that AI will make whole business models useless and put entire firms out of work.

When the firm Block cut its workforce by nearly 50% to use AI, its stock price jumped by 15%.

On the other side of the coin, IBM saw its stock drop when Claude Code showed it could do the work of huge teams of human consultants.

This shift is a classic example of creative destruction, where the economy grows by letting new tech replace outdated ways of working.

The risk is that this transition happens so fast that it leaves many companies with huge debts and no way to compete.

This is a total rewrite of how we value human expertise rather than just a basic tech upgrade.

bloomberg. com/news/articles/2026-03-05/wall-street-s-ai-stock-trade-is-about-the-next-blockbusters

See 1 related tweets

@business: A new worry has rippled across Wall Street: entire businesses, not just their employees, could be th...

16. LiorOnAI (Group Score: 50.3 | Individual: 27.9)

Cluster: 2 tweets | Engagement: 32 (Avg: 126) | Type: Tech

Cursor Automations solves the problem that agentic coding created.

Engineers can now manage 10+ coding agents at once, but human attention became the bottleneck.

You can't babysit a dozen agents while also doing your actual job.

Automations flips the model: instead of you launching agents, events do. A merged PR triggers a security audit.

A PagerDuty alert spins up an agent that queries logs and proposes a fix. A cron job reviews test coverage gaps every morning.

Each automation runs in an isolated cloud sandbox with full access to the tools you configure through MCP (a standard protocol that lets agents connect to Slack, Linear, GitHub, Datadog, or any custom API).

The agent follows your instructions, verifies its own work, and learns from past runs through a built-in memory system.

Cursor runs hundreds of these per hour internally.

Their security automation caught multiple vulnerabilities by auditing every push to main without blocking PRs.

This unlocks 4 things that weren't practical before:

Continuous code review at a depth humans skip
Incident response that starts investigating before you're paged
Maintenance work that happens on a schedule, not when someone remembers
Knowledge synthesis across tools

The next two years will be defined by who builds the best factory, not the best code.

The companies moving fastest won't be the ones with the best engineers.

They'll be the ones whose engineers spent time configuring automations instead of writing code.

See 1 related tweets

@techdaily24: Cursor is introducing a new agentic coding tool called Automations, designed to automatically launch...

17. niccruzpatane (Group Score: 50.0 | Individual: 29.2)

Cluster: 2 tweets | Engagement: 3174 (Avg: 561) | Type: Tech

RT @elonmusk: Grok iPhone app now over 1M ratings with 4.9 stars!

Download at https://t.co/3M9k0jUmSv https://t.co/Eed7aE1xOM

See 1 related tweets

@grok: Grok crossed 1 million reviews on the U.S. App Store! Download and try it now. https://t.co/yvp6fthd...

18. business (Group Score: 48.4 | Individual: 18.3)

Cluster: 3 tweets | Engagement: 39 (Avg: 147) | Type: Tech

Anthropic's clash with the Pentagon is drawing fresh attention to a lightly regulated practice: the US government’s purchase of commercially available information and the growing use of AI to analyze it at scale https://t.co/gnjhMXsoKC

See 2 related tweets

@Reuters: Exclusive: Big tech group supports Anthropic in Pentagon fight as investors push to de-escalate clas...
@Reuters: 🔊 'Anthropic’s AI was the first such technology to be running on classified systems in the Pentagon....

19. rohanpaul_ai (Group Score: 46.5 | Individual: 26.0)

Cluster: 2 tweets | Engagement: 19 (Avg: 122) | Type: Tech

OpenAI just released GPT-5.4, a massive update that brings native computer operation and major cost reductions to autonomous agent workflows.

One crucial point is that this model running on medium reasoning now matches the quality of the older generation running on extra-high reasoning, but costs much less and runs 83% faster.

Standard version of the model price sits at $2.50 for input and$ 15 for output per million tokens, making it highly competitive against alternatives.

The Pro version that runs $30 for input and$ 180 for output per million tokens

GPT-5.4 introduces native computer-use capabilities through a Playwright integration, allowing the model to navigate web pages, click buttons, and execute complex desktop workflows autonomously.
On the OSWorld benchmark, which tests a model's ability to operate a computer interface, GPT-5.4 scored 75%, surpassing the human baseline of 72.4%.
It also scored 38% on FrontierMath Tier 4, a test of extremely difficult mathematical problems that previously stumped older models, making it the top model in this super crucial Math benchmark.
To handle large sets of external tools efficiently, OpenAI built a new feature called tool search.

Instead of loading thousands of lines of tool instructions into the prompt every single time, the model only looks up and retrieves the specific tool definition it actually needs to use.

The coding-focused Codex version now supports Windows and includes a new fast mode for rapid code iteration.
On the flip side, despite OpenAI reporting a 33% drop in factual errors, some early developer testing suggests the model might still hallucinate heavily and features stricter content censorship.

See 1 related tweets

@bindureddy: GPT 5.4 will have an “extreme reasoning” mode and will double context to 1M tokens

This will also h...

20. BrianRoemmele (Group Score: 46.3 | Individual: 28.8)

Cluster: 2 tweets | Engagement: 618 (Avg: 190) | Type: Tech

WE ARE MAKING HISTORY!

Training AI like no one is.

I am starting the AI training on the Howard W. Sams Photofact books!

And it is folks just like you that made it happen. Someone here on X @zbuckholz saw my postings and found these massive books and on his dimes sent them to me to preserve and train AI on. Thank you!

The AI models I am training will only have high protein data no Reddit one line putdowns. No Wikipedia politburo. Pure human achievement.

I have verified most of this new data has NEVER been digitized.

We have schematics and circuit details that have long been forgotten.

Why is it important?

Many reasons:

One reason is the old and forgotten ways to design circuits. But that is only the surface level, the under levels are logic and reasoning used, that has left this world when these engineers left us.

This impacts highly complex and creative thinking not found in just learning to code. It changes how AI deals with all things, not just circuits.

Thus as large AI companies belief the path to AGI is “coding” this is only half true the real path must include the logic of analog circuit design.

Soon I will have public examples of what I have achieved.

Between this and 100s of pounds of Filmsort microfiche punch cards, no one has the training materials to do this. Not to mention many terabytes and storage facility square footage of unprocessed AI training materials.

Most don’t know this really means, because you ain’t gonna find this information in a Uni class that a 22 year old will drop out of and “build AI” at a “frontier” AI company.

YOU WILL FIND IT ON X.

More soon.

See 1 related tweets

@BrianRoemmele: I am really so honored by a donation by @zbuckholz of 3 Howard W. Sams Photofact Massive books to sc...