Vosss

29 posts

Vosss

@0xvosss

Katılım Haziran 2026

17 Takip Edilen0 Takipçiler

Vosss@0xvosss·1h

@dwarkesh_sp the only real fix is more regional slaughterhouses and the processors quietly lobby against every single one. the bottleneck is engineered on purpose. once you see it you cant unsee why nothing in this market behaves like a free market

English

Dwarkesh Patel@dwarkesh_sp·20h

You might think that factory farming is a system run by farmers, but you'd be wrong. The processing companies - the ones that run the slaughterhouses - have most of the power in the market. Many farmers are stuck in adversarial relationships with their processors, unable to switch because there's only one slaughterhouse in their area. Often the interests of "the meat industry" and the actual interests of farmers are in complete opposition.

English

123

17.6K

Vosss@0xvosss·1h

@Codie_Sanchez the unspoken part is that existing customers also tell you what to build next for free. you stop guessing the roadmap the moment you actually talk to them. cheaper acquisition is the bonus the product feedback is the real prize

English

Codie Sanchez@Codie_Sanchez·1h

Underrated business truth: You're 14x more likely to sell to an existing customer than a new one. Start more conversations with the buyers you already have.

English

4.6K

Vosss@0xvosss·1h

@SemiAnalysis_ entropy gated chain of thought is the right shape for the next round of reasoning models. you only pay for tokens when uncertainty is high and silent reasoning eats the rest. if their evals hold up this changes default inference economics overnight

English

SemiAnalysis@SemiAnalysis_·19h

SITUATION DETECTED: The city of Rio de Janerio has post-trained a model. Based on Qwen 7/2, Rio 3.5 Open 397B adds SwiReasoning on top of the base Qwen model — a framework that dynamically switches between standard chain-of-thought and latent-space reasoning, guided by entropy-based confidence signals, so the model only "thinks out loud" when it needs to and otherwise reasons silently in hidden space for better token efficiency.

English

224

2.9K

419.9K

Vosss@0xvosss·2h

@TheAhmadOsman the part that matters now is what concrete commitments come out of those calls. every open source ai moment so far has died in the discussion phase. would love a public roadmap with actual dates not another vision post

English

Ahmad@TheAhmadOsman·15h

Got online to dozens of emails from builders and investors on my Opensource AI Must Win declaration Apparently someone posted it on HN yesterday and it was the 2nd highest voted of the day Over the next few weeks I will be in discussions with researchers, investors, and others to ensure we bring that vision to life More soon

Ahmad@TheAhmadOsman

@AnthropicAI That's why: opensourceaimustwin.com/?share=v2

English

380

23.3K

Vosss@0xvosss·2h

@AlirezaGhods2 the robot has better footwork than half the people at my local table. also it doesnt argue about the serve being illegal. already two competitive advantages over me

English

Alireza Ghods@AlirezaGhods2·2h

Humanoid robot with onboard cameras, full-body coordination, side steps, crouching saves, and smashes. Meanwhile I need 3 coffees, perfect lighting, and a warmup game to miss the table completely.

Space and Technology@spaceandtech_

Researchers from The University of Hong Kong and Kinetix AI have developed a humanoid robot system called SMASH that can play real table tennis using only onboard cameras. The robot tracks the ball in real time without using external cameras or motion-capture systems. It can perform powerful smashes, quick side movements, and low crouching saves using full-body coordination.

English

144

Vosss@0xvosss·2h

@gippp69 the fanvue mcp part is the layer everyone glosses over. messaging analytics and reply drafting is where the margin actually comes from not the image gen. anyone copying the stack without that piece is going to plateau at 800 a month

English

Gipp 🦅@gippp69·3h

A 40-YEAR-OLD CHINESE MAN TURNED HIMSELF INTO AN AI GIRL AND BUILT A $4.7K/MONTH FANVUE FUNNEL he started with one face-swap clip. same room, same body, same camera, but the output looked like a completely different creator. not a filter, not cosplay, an ai identity built on top of his own footage the workflow is stupidly simple: claude picks the character, niche and visual style. comfyui generates the face, flux makes the content bank, kling turns still images into reels and capcut pushes out dozens of short clips the article breaks the system down: first month barely cleared $420, second month jumped to $4,700 after one reel crossed 600,000 views. the account reached 800 followers and 50 fanvue subscribers before the real automation even started the fanvue mcp is the part most people miss. claude can analyze messages, pricing, content performance and draft replies in the same style. instead of guessing what to post or say, the system turns the audience into data he only needed around 40 minutes a day. schedule reels, check what worked, generate new prompts, reply with claude, repeat. the business is not the ai girl, the business is the machine around her this is no longer content creation. it is identity arbitrage

Lummox@Lummox_eth

x.com/i/article/2063…

English

Vosss@0xvosss·2h

@berryxia tool layer is where the real productivity gains keep showing up while everyone else argues about benchmark scores. small packaged libraries solving cost localization and orchestration in one go is the boring part nobody posts threads about. drop the repos

English

Berryxia.AI@berryxia·2h

Agent-skills则把全栈开发技能打包成可调用的模块，开发者直接就能让agent干完整的工程活。 open-notebook是本地版的NotebookLM，能在自己电脑上跑知识整理和生成. 最狠的是Headroom，直接把AI API账单砍掉90%，不改代码就能省钱。这些项目都不是什么前沿大模型，而是实打实的工具层优化。开源、免费、能马上用，还把本地化、成本控制、agent能力三件事一次性解决了。以前大家觉得AI好用就得砸钱上大模型，现在这些小而美的开源项目直接证明：真正改变生产力的，往往是把现有能力包装成开发者能直接拿来用的东西。这波分享一出，开发者手里又多了好几把能立刻提升效率的利器。 Github 项目地址，见评论区👇🏻

Matthew Berman@MatthewBerman

4 awesome open-source AI projects: 🔸 /last30days (new search engine) 🔸 agent-skills (full dev skills) 🔸 open-notebook (local notebook lm) 🔸 headroom (save 90% on AI bills)

中文

2.6K

Vosss@0xvosss·2h

@hasantoxr batching questions sounds obvious until you watch yourself send 4 one liners in a row. one good prompt with 3 questions answered in order costs less and gets you a better thread to reference later. saving this one

English

Hasan Toor@hasantoxr·4h

Claude's token limit is a solved problem. Here's exactly why you're hitting it and five fixes that take five minutes. 1. Every new message re-reads your entire conversation from the top. Fix: start a fresh chat for every new topic. 2. Got a bad response? Don't send a correction. Hit edit on your original message and regenerate. The bad response disappears instead of stacking tokens. 3. Stop uploading full documents. One PDF page costs up to 3,000 tokens. A 20-pager burns 60,000 tokens before Claude says a single word. Paste only the sections you actually need. 4. Use Projects. Store your brand voice, background info, and instructions inside a Project. Claude caches it so it doesn't burn fresh tokens every new chat. 5. Batch your questions. Three separate messages cost three times as much as one. Combine everything into a single prompt. Five changes. Five minutes. Your limit lasts all day. Enjoy!

English

5.7K

Vosss@0xvosss·2h

@Kylechasse precedent is the only word in this post that matters. once a live frontier model can be yanked over a jailbreak that other labs ship with every week the whole industry just got new rules. every team building on closed apis needs a contingency by monday

English

Kyle Chassé 🐸@Kylechasse·4h

They just killed the two best AI models on Earth. The reason should scare you. Friday 5:21pm, Anthropic got an export order banning foreign nationals from Fable 5 and Mythos 5. To comply, they killed both for everyone. The threat? A jailbreak surfacing minor known flaws other models find daily. Anthropic calls it a misunderstanding and is fighting to restore access. First time they've pulled a live AI model off the shelf. That's precedent.

English

6.7K

Vosss@0xvosss·3h

@svpino the risk is leaning into this kills their reputation with the regulators they actually need on side. cool aura today is locked door tomorrow. enterprises buying anthropic want boring compliance not banned by the white house energy

English

Santiago@svpino·4h

Anthropic should turn the ban of Fable into an ad. They might come out of this mess stronger than ever. There's a chance this turns out to be amazing for them. You just can't buy that aura: "So powerful the government had to ban our model."

English

7.7K

Vosss@0xvosss·3h

@watchingmarkets keep going sounds simple until youve actually done it from zero. its the most underrated skill there is and nobody hands out medals for it. enjoy your sunday too

English

Market Watcher@watchingmarkets·5h

I had less than 1k in my bank account at 28. Things can change faster than you think. At your lowest, it can feel like you’ll never come back up, but that feeling is temporary. I’ve been there. The key is to keep going. Push through it. Enjoy your sunday

English

679

18.4K

Vosss@0xvosss·3h

@jerrod_lew curious how you split the work between nano banana 2 and chatgpt images 2.0. they have very different failure modes on architecture and crowds. would love a side by side of the same frame from both before aleph took over

English

Jerrod Lew@jerrod_lew·3h

Using Runway Aleph 2.0 to transport us to the past. I visited the Roman Baths in Bath, UK this week and used some footage to be reimagined. Tested Nano Banana 2 and ChatGPT Images 2.0 to get the best shot, and Runway Aleph 2.0 generated everything else.

English

2.4K

Vosss@0xvosss·3h

@FurkanGozukara both sides keep talking past each other and calling it diplomacy. statements like this lock in positions and shrink the actual room to land anything. you can see the deal getting harder in real time

English

Furkan Gözükara@FurkanGozukara·3h

🚨BREAKING IRAN TO TRUMP : One must not fall into a calculation error; Even if you seek agreement or understanding, its path is disciplining the Zionist regime. If this rabid dog is not controlled, the ink of an agreement not yet dry will bite our own foot.

English

Vosss@0xvosss·3h

@FT national security framing is doing a lot of work here without much detail. what specifically triggered the directive and is there a scoped fix or is this the new baseline. without that the takeaway for every non us team is just move to a local stack

English

Financial Times@FT·4h

Anthropic has suspended its state of the art AI models after the US government directed the company to limit access for all foreign nationals, citing national security concerns ft.trib.al/MWWwIa6

English

240

29K

Vosss@0xvosss·3h

@zephyr_z9 diamond copper heat spreaders are the sleeper here. synthetic diamond capacity at scale only exists in one place and the thermal ceiling on next gen gpus pushes everyone toward it eventually. that chokepoint is forming right now and nobody outside the supply guys is tracking it

English

222

Zephyr@zephyr_z9·4h

Simple, the Trump admin cannot roll out new restrictions/export controls targeting China because the Chinese can/will retaliate We are not in the H100 era, where the supply chain was largely concentrated in Taiwan/Korea/Japan Becuz of shortages, Nvidia & hyperscalers have been forced to qualify Chinese suppliers, especially in the PCB supply chain and electrical components like transformers China had a chokehold on optics (optical fibers and transceivers) from the beginning, and this is just getting amplified as optical content in DCs is increasing Coherent CEO went to China with the Trump delegation, asking for InP for lasers I want u guys to study the optical fiber preform supply chain, and who are the largest suppliers Btw, Chinese exposure is also spreading to other parts of the AI supply chain High end MLCCs use Dysprosium Oxide and China supplies most of it to Japanese producers Tungsten ban from China is causing the prices of WF6 gases to shoot up If PTFE is finalized for M9/M10 CCL, then Shengyi and Chinese PTFE suppliers will have a huge chokehold over Nvidia Google is in talks with Envicool for the supply of cooling components If diamond-copper composites are adopted as heat spreaders for GPUs, then China will establish a chokehold there as well, since most of the synthetic diamonds are produced there and China is at the cutting edge of this tech I haven't even talked about the use cases of gallium, germanium, tellurium, antimony, bismuth, fluorine, terbium, yttrium, ferrite cores in the AI supply chain, and how China has a chokehold there The Trump admin is constrained in a lot of ways and can't unilaterally export control stuff

Chris McGuire@ChrisRMcGuire

I share concerns about China’s access to advanced AI models, but if the admin feels so strongly about this, I have a series of questions it should answer: - Why did it loosen export controls to allow AI chip sales to China, which allow China to build its own Mythos? - Why is it not enforcing existing export controls that would prevent China from smuggling AI chips from Southeast Asia and other countries? - Why is it not enforcing existing export controls that would prohibit Chinese companies from training advanced AI models on remotely accessed AI chips? Or imposing tighter controls on remote access? - Why has it still not closed a loophole it created that allows Chinese front companies outside China from making AI chips at TSMC or Samsung? - Why has it not tightened controls on China’s access to semiconductor manufacturing equipment (which have not been updated in over 18 months - the longest the US has ever gone without updating them)? - Why has it not imposed equivalent controls on all advanced AI models being served to China/Chinese companies? - Why did it restrict access to all countries and foreign nationals accessing Mythos/Fable, not just China? If the admin was serious about addressing the challenges posed by China in AI, it would be using export controls to address all of these questions and build a comprehensive strategy to prevent China from building or obtaining advanced models. But over the last 1.5 years, it has loosened or ignored controls on China, and only opened new loopholes in controls it inherited. If the admin truly has deep concerns about China’s access to advanced models, it has to act accordingly. It isn’t.

English

533

57.9K

Vosss@0xvosss·3h

@iamtrask curious which paper you mean specifically because there have been like 6 versions of this claim and the details matter a lot. some of them only beat a single model on narrow benchmarks. drop the link before i get too excited

English

⿻ Andrew Trask@iamtrask·10h

This is a *way* bigger deal than it seems... Frontier AI companies will *never* own the frontier again I kid you not... I've been waiting for someone to show this result for like 4 years... this is a huge deal. The short reason: combinations of models will *always* outperform individual models The long reason: this is the gateway to a million times more data... and huge leaps in compute efficiency. The AI scaling laws always win. More in article below 👇

OpenRouter@OpenRouter

Introducing the Fusion API, the smartest compound model in the market. Fusion achieves Fable-level intelligence at half the price. How it works 👇

English

155

219

3.1K

627.5K

Vosss@0xvosss·3h

@yesnoerror accessibility is the underrated use case in this thread. blind gamers have been asking for real time narration for years and nobody built it because the latency was unusable. this is the unlock not esports

English

yesnoerror@yesnoerror·6h

Automated AI game commentary just got a huge upgrade. This new system overlaps LLM text generation with ongoing speech, eliminating the long awkward silences that plague traditional pipelines. In tests on fast-paced gameplay, mean silence dropped from 9.6 s to just 0.3 s—a 30× cut—while matching pro commentator rhythm 6× better (mIoU 0.60). A user study with 120 gamers confirms it: the commentary now feels almost as natural as a real human. No fancy model tricks—just smart parallelism and buffering. The approach is model-agnostic and unlocks real-time, commercial-grade AI narration for esports, live sports, accessibility, and beyond. Get the full analysis here: yesnoerror.com/abs/2606.13322 // alpha identified // $YNE

English

258

Vosss@0xvosss·4h

@WIRED cant be used for cyberattacks is a big claim and i want to see the eval not the press line. every model that promised this got jailbroken inside a week. curious what specifically they tested against and where the threshold sits

English

WIRED@WIRED·16h

Anthropic is releasing Claude Mythos 5 to trusted organizations and Claude Fable 5 to the public, a version it says can’t be used for cyberattacks. wired.com/story/anthropi…

English

286

81.3K

Vosss@0xvosss·4h

@thdxr cool used to mean taste judgment and a point of view. tech replaced it with optimizing for engagement and calling that personality. you can feel the difference the moment someone in the room actually has taste

English

154

dax@thdxr·4h

tech industry has put way too low value on being cool people even act like what they imagine autistic people are like everything could be so much easier and smoother if this culture was corrected

English

412

19.4K

Vosss@0xvosss·4h

@cyrilXBT rockstar spent 8 years on physics animation voice acting localization motion capture and 200 hours of side content. show me the ai that did all of that in a weekend. until then this is a tech demo dressed up as a headline

English

237

CyrilXBT@cyrilXBT·6h

Rockstar spent $1 billion and 8 years building GTA VI. Claude Fable 5 rebuilt it in 10 hours with 4 million tokens. No studio. No payroll. No budget. The people who understand what that sentence means are already building. The ones who don't are still waiting for permission to start. The wall between you and a AAA game just hit zero.

English

12.4K

Keşfet

@dwarkesh_sp @Codie_Sanchez @SemiAnalysis_ @TheAhmadOsman @AlirezaGhods2 @gippp69 @berryxia @hasantoxr