AI 行业热点

🐦 X/Twitter 热点

Josh Woodward (@joshwoodward)

  • Big milestone! [235 ❤️ 9 🔄]

Boris Cherny (@bcherny)

  • Opus made an infographic [117 ❤️ 1 🔄]
  • I used Opus 5.5 to formally verify the Claude Agent SDK using Lean. A couple short prompts = 16 PRs fixing various bugs and race conditions. Video attached.

TLA+ also works well. I sometimes combine Lean and TLA+ to look for issues around data flow, concurrency, and state mgmt.

I don’t know either language well, but Claude is excellent at both. This approach is super useful for formally modeling your code and finding bugs that a human probably wouldn’t have spotted.

Is formal verification the future of coding (or at least, bug finding)? [3324 ❤️ 165 🔄]

  • Opus 5.5 is a really good model. It’s been my daily driver the last few weeks.

We had Opus 5.5 and Fable 5.1 each port HAProxy from C to Rust. Both passed nearly all of HAProxy’s tests, but Opus 5.5 finished in 9.5 hours compared to Fable 5.1’s 12 hours, and for 51% less cost. [6778 ❤️ 228 🔄]

Thibault Sottiaux (@thsottiaux)

  • Maybe our cutest launch so far. But still packing the biggest punch. [6045 ❤️ 134 🔄]
  • GPT-6 Sol and Luna are out. Not only are they a very significant improvement across the board, but also in writing and general “you know when you try it” quality.

We are also permanently reducing the API price by 50% making both of them viable for a ton of new usecases and making your usage go further too, even on the subscriptions.

And one more thing. We are loading a banked reset into all accounts of our Plus, Pro and Business users. Let’s go! [23309 ❤️ 1512 🔄]

  • We have been focusing on efficiency and intelligence for all.

Very proud of the team. Only possible when you have incredible models at the top end of the capability that you can then use to make a big difference in everything else. [12590 ❤️ 317 🔄]

Peter Yang (@petergyang)

  • Just randomly played this Australian show on Apple TV and it’s another banger [19 ❤️]
  • @claudeai If you enjoyed this, sign up for free to my newsletter to get my best AI and product guides.

Join 110,000+ subscribers here: [1 ❤️]

  • I talk about why Opus 5.5 is the most excited I’ve been about a Claude model in awhile here: [3 ❤️]

Cat Wu (@_catwu)

  • If you’re looking for ways to try Opus 5.5 and Claude Tag in Slack… [119 ❤️ 2 🔄]
  • Claude Opus 5.5 is now the default model in Claude Code and the Claude app, including Cowork, for Pro, Max, and Team plans. I’ve been using Opus 5.5 as a daily driver and love its clear communication skills and ability to write in my style.

We’re defaulting to effort medium across products, which is comparable to Fable 5.1 on intelligence but faster. Your rate limits will go 25% further on Opus 5.5 compared to Opus 5.

Give Opus 5.5 an ambitious task and let us know what you think! [1308 ❤️ 35 🔄]

Thariq (@trq212)

  • if you want to make games, 3d generation is a great capability to help you imagine your game come to life

but you should figure out how to make a good, satisfying game loop first

(talking to myself here) [254 ❤️ 5 🔄]

  • the right way to use model capabilities is not to ship 10x more features to prod

it’s to spend more time understanding your users, trying experiments, building prototypes, learning about things you don’t understand so that you can ship things that actually work [4528 ❤️ 313 🔄]

  • workflows are a huge part of how I use Claude these days, I’m glad we have Fable-like intelligence that works with workflows from a cost perspective [73 ❤️]

Amjad Masad (@amasad)

  • Your intelligence is fake. It’s artificial.

Mine? It’s super. Super Intelligence. Aka SI. [935 ❤️ 37 🔄]

Guillermo Rauch (@rauchg)

  • Software will never die again. You liked Google Reader? Cool, you can generate and deploy your own. Yours, forever. [1045 ❤️ 33 🔄]
  • We ran fresh Next.js evals. The tally:
    ① Opus 5.5 [𝟿𝟽%]
    ② GPT 6 Sol [𝟿𝟽%]
    ③ Fable 5.1 [𝟿𝟽%]
    ④ Grok 4.7 [𝟿𝟺%]

Notably, Grok is 2x-7x cheaper [1087 ❤️ 55 🔄]

  • Love the tastefulness with which anthropic ships.

Main reason I got excited about ‘headless’ for the web is that any given page could take whatever whimsical or unique shape you wanted. With AI, there’s no excuse not to push the design frontier.

This is @nextjs at its finest: [1418 ❤️ 34 🔄]

Alex Albert (@alexalbert__)

  • Prompt in Claude Tag

“Recreate Market Street, San Francisco as it stood on April 17, 1906, the afternoon before the earthquake, in Blender.

Scope: the Ferry Building up Market to Fifth Street, including the Palace Hotel, the Call Building, the Chronicle Building, Lotta’s Fountain and the Emporium.

Before modeling anything, build a source file from: the 1899-1905 Sanborn fire insurance maps (footprints, heights, materials, occupants), the Miles Brothers film “A Trip Down Market Street” (April 1906), period photographs from OpenSFHistory, the Library of Congress and the David Rumsey collection, and USGS topography. Record every building with footprint, height, facade material, occupant, and the source for each fact with a confidence level.

Build everything in Blender Python. No downloaded meshes, textures or HDRIs. Write reusable generators (Victorian commercial facade, mansard roof, bay windows, awnings, painted signage, gas and electric street lamps, cable car, horse-drawn wagon, early automobile) and assemble the street from the source file, so every building traces back to data.

Provide a 10 second video up the street.” [73 ❤️ 1 🔄]

  • I’ve been on a Blender kick with Opus 5.5. Its better 3D modeling and vision mean you can build an entire world from a single prompt.

Historically accurate San Francisco Market street in 1906, pre-earthquake [890 ❤️ 40 🔄]

  • need to make sure you’ve been updated to this (rolling out to everyone soon if you haven’t!) [31 ❤️]

Aaron Levie (@levie)

  • What an insane day in AI. The frontier models just became substantially cheaper, with the Opus 5.5 price cuts, and now with GPT-6 Sol and Luna dropping token prices by 50%.

The rate at which the cost per task (on a like-for-like basis) drops in AI is unlike any other type of technology in history. And every time the cost of AI drops, the use-cases you can deploy agents against dramatically increase. This is Jevons paradox applied to agents.

These improvements will directly lead to broader diffusion of AI in the economy as we can use agents to process all of our data, scan our code for security issues, read through all log data to make decisions, have agent swarms in workflows, and much more. The cost of tokens is directly correlated to these use-cases being opened up at scale. [655 ❤️ 69 🔄]

  • At Box, we’ve been testing Opus 5.5 on a variety of complex enterprise knowledge work tasks dealing with unstructured data with the Box Agent.

Overall, we saw frontier capability levels, with major performance improvements over Opus 5. 63% fewer tokens used, 42% less verbosity, and 30% faster vs. Opus 5. And the model itself is cheaper, so this is a major win for any agentic computer use, coding, analytics, or data work that enterprises will be doing.

Here are some examples of the task wins and performance gains across a variety of industry tests that we performed:

• Financial services - due diligence (+39% task accuracy): A year of transaction records, with the job of finding every miscalculation in an acquisition target’s pricing tool. Opus 5.5 scored a perfect result on every attempt in half the words Opus 5 used, consuming 82% fewer tokens overall.

• Technology - cloud cost analysis (+65% task accuracy): Work out what a company should actually change about its cloud spend. Opus 5.5 picked the right basis for the retention calculation and kept the source data’s unit conventions straight all the way through, so the number at the end actually holds up. It took half the time Opus 5 took, with 70% fewer tokens.

• Consumer products - client account analysis (+17% task accuracy): Set the onboarding targets for a client account, reading across the signed contract, a satisfaction tracker and a team metrics sheet. The contract never states a senior/junior split, so Opus 5.5 derived it from the 18-person roster and showed the rule it used; several clients had a perfect 10 on individual survey questions, so it averaged each client’s responses instead of crowning the single 10. It finished this one in half the time, on 78% fewer tokens.

• Clinical diagnostics - data analysis (+15% task accuracy): Malaria rapid-test performance across a dry and a wet season: build the patient records out of two clinical PDFs, compute positive test rates by season and gender, and test whether parasite counts really differ between test-positive and test-negative patients. Opus 5.5 caught that the two groups’ standard deviations differed more than 100-fold, re-ran it the right way, and found the dry-season difference didn’t hold up after all. This accuracy gain came with a final answer that was half the length of Opus 5’s, and also needed 78% fewer tokens end to end.

Customers will be able to build AI Agents with Opus 5.5 shortly in the Box AI Studio. [232 ❤️ 23 🔄]

Garry Tan (@garrytan)

  • Capy (@capydotai) lets me drop PRs much much faster than I would with Codex or Claude Code alone [111 ❤️ 9 🔄]
  • If you’re working on something cool, GStack does actually tell you to apply to YC :-) [155 ❤️ 3 🔄]
  • We need to teach the world to prompt and maximally use AI: to help all people see the way it can give you wings in all your pursuits

Then we need yearn to solve more problems for ourselves and for one another [144 ❤️ 17 🔄]

Nikunj Kothari (@nikunj)

  • The amount of SPVs organzied by even the “best” investors is simply mind boggling..

To the point that if you read headlines of a large fundraise, you can be reasonably sure now that a large amount of it came from SPVs even though the headlines say something else.

Combine that with tranched valuations and revenue not being the same, all I’ll say don’t believe the headlines anymore.

Proof is in the pudding and the pudding is not publicly visible anywhere. [34 ❤️]

  • Now also on YouTube:
  • A Walk In The Park (part III) feat @toddsaunders

00:00 Software should fit your business
00:16 Meet Todd, the councilman
03:01 From Google to flooring software
07:51 The $10M pivot
11:55 Life after the exit
17:35 Vertical vs. horizontal software
23:21 The college startup story
24:35 Cofounders and the Dalton Mills name
28:06 Why raise VC again?
32:43 Who Todd wants to hear from

shot on @iPhone, edited on @descript
📍westfield, nj [34 ❤️ 2 🔄]

Peter Steinberger (@steipete)

  • Had ChatGPT sometimes crashing on me after updating to macOS 27 and… Astra found a ~14 year old bug in libuv. [850 ❤️ 33 🔄]

Dan Shipper (@danshipper)

  • @every
  • i added almost 10k new followers today!

if you’re new around here, here are two things to start with:

  1. my vibe check on opus 5.5 vs. sol-6 on @every:
  2. after automation, my take on why AI automation creates more good work to do for human experts:

if you like those two things you should subscribe to @every—there’s more where that came from! [37 ❤️ 1 🔄]

  • 😚 [16 ❤️]

Aditya Agarwal (@adityaag)

  • I have hosted over 50 Minus One fireside chats with lots of interesting and well known people over the years @spc.

The first one my 9 year old asked to attend was yesterday’s with Dmitri of Waymo.

The evocative beauty of building things for the physical world is undeniable. [17 ❤️]

  • Safety and alignment of “models” is the raging topic today.

But the original “safety” debate for AI/MachineLearning was actually in autonomous vehicles!

My first Waymo ride truly felt like a religious experience. It was mind-boggling that we could build technology that actually worked on real roads with all the challenges that entails.

I thought the most interesting part of the talk yesterday was the extensive eval/testing infrastructure that Waymo has built out in order to build the confidence to release 2-ton robots traveling at 30mph through urban environments.

It was awesome to host @dmitri_dolgov @spc yesterday.

I learnt a lot myself and I am excited to release the full video soon. [41 ❤️ 4 🔄]

  • In the mad dash for productivity…we often forget that AI products should be delightful and FUN.

I have had so many magic moments of delight on Sentience.

Highly recommend trying it out. [23 ❤️ 1 🔄]

Sam Altman (@sama)

  • argh i meant VOICE not video :(

sorry to dissapoint [855 ❤️ 16 🔄]

  • Startups are naturally good at this; it is hard to keep a bigger company good at this and i think an underexplored space. [3675 ❤️ 123 🔄]
  • michelle embodies this as much as anyone

openai is so, so lucky to have benefited from everything she has done so far and i think people will be quite pleased to see what she + team have cooking next! [2124 ❤️ 58 🔄]

Claude (@claudeai)

  • Claude Opus 5.5 is available today. What will you explore? [226 ❤️ 5 🔄]
  • An algorithmic drawing program created by @kevin_t_ngo. Every drawing is based on a different seed. [261 ❤️ 7 🔄]
  • A toy brick-building app. Turn a photo or a description into a buildable model with custom instructions. [257 ❤️ 7 🔄]

由 Follow Builders 自动生成 · 2026-09-23