Samar Singh

वनमाली गदी शार्ङ्गी शङ्खी चक्री च नन्दकी श्रीमान् नारायणो विष्णुर्वासुदेवोऽभिरक्षतु।

@miladreaamy 35kg

Bunu çözersen, sen bir dahisin. Garfield'in kilosu nedir?

@AI_evangelist42 Opencode is not as good as CC

turns out the Claude Code harness sucks. the only reason Claude Code is so great is because of the Claude models, not the harness

@actionxander Ooh !!! Tat sinter can break anytime and they will be deep inside scalding superheated water

Some tourists walked out onto Grand Primsatic at Yellowstone National Park. Unbelievable.

@puravmanot @heyclicky What they built is bits and pieces taken from real opensourced contributions from @NousResearch etc.

I've found issues in @heyclicky's binary re: data collection and potential discrepencies with the privacy policy. HeyClicky 1.0.40 (build 49) dmg SHA-256: 39fd09f7e029d5c51f154c72bbb2b6925583848e2f23e509b52140f5c08a1589 1/9 (a thread)

@rodamn I have model X, staggered setup consisting of 265/35R22 on the front wheels and 285/35R22

@samarknowsit Ouch. Which wheel set do you have

If all car companies were like Tesla, people would be so much less disgruntled Wife drove off with a flat tire (20 psi), called me I tell her to bring it back home and to take my car Whip out my phone, Tesla app, roadside help, select 1:00-1:30 appointment to get her tire replaced Literally seconds to get the process started They just showed up at my driveway, removed the offending tire, found a screw in the tire, installed a loaner wheel/tire Tech will be back later today to install her repaired wheel and tire A few taps on the phone, two minutes on the phone, and a text message to get this done How does it get any better than this?

@Pivot2Centre Kenya, El Salvador, Indonesia, Philippines, all have had these. Just FYI

India does 1st world stuff such as this and then India can’t solve 3rd world problems such a city roads, urban drainage, and urban garbage mgmt!

@deanwball Very idiotic post TBH, and wtf is head of strategic futures

Some observations on Kimi: 1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run. 2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China. 3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex. 4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business. 5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this. 6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

@nikesharora Superb

In the presence of legends! This is not an AI post :). But for those who know, they know.

@HealthRanger And takes all your data and feeds into Chinese propaganda

David Sacks is correct here. I've been using Kimi-K3 all day today. It's the OPPOSITE of woke. It doesn't lecture you. It doesn't censor you. It just works for you, on almost anything you want. And it's VERY good at finding and fixing code bugs.

@DavidSacks True ! First time, I agree with what David says. There is no need to lobotomize capabilities

OH: “i’ve switched to Kimi from claude for a bunch of work. it’s just so much more fun because it just does the thing instead of lecturing you” Woke lobotomized models are the enemy of American competitiveness.

@mukund Because India doesnt have any opportunities! Its all about food delivery, e-commerce and babugiri. Nothing else

Why do you think so many Chinese go back to China after education in the US but not many Indians do or want to ? No cliches please.

@SuperChilligan @HarryStebbings Starbucks recently pulled an AI-powered inventory counting system after it produced inaccurate counts, forcing stores to revert to manual tallies. So goodluck with getting this done

@HarryStebbings This is becoming a trend. For instance, Starbucks is shifting away from enterprise software vendors like Microsoft, IBM, and Oracle to cut its $400 million annual software budget.

"We replaced Salesforce with a vibe-coded CRM built for our own workflows. The custom system integrated our AI agents more effectively, worked better for the team, and made Salesforce unnecessary. That decision cut a $600,000 annual software bill to zero." Is this an anomaly or the start of a much larger trend @chamath @Avishai_ab @jasonlk @benioff

These are all BS stories, for millions of customers this kind of system wont scale nor will be the agent reliable. There is literally no example where someone took an agentic autonomous system and deployed it in prod. In almost all cases LLMs removed fractional effort resulting in somewhat improved flow, but couldn’t get the cost factor completely out.

@denisewu I think this is carry forward from prior training data.

There’s a large grain of salt!🧂 Kimi is still calling itself Claude.

@bradmillscan Or jusy install openvpn on a small Ubuntu box

Going on a trip and need access to your Agents on your home server, but you want to keep FileVault & other security measures? UPS <> Smart plug <> Your Mac <> JetKVM connected to tailscale boom, access from anywhere

@AlexKolicich Idiot

It looks likely that Kimi K3 was heavily distilled from Fable If that can be proven, the U.S. government should act aggressively against the theft of American IP It cannot ban Fable for a month, then do nothing when a foreign competitor releases a distilled copy

@Prigoose This is absolutely disastrous!

what a live ceo looks like

@DogukanUrker This is old trick ! Read Kimi’s paper

Half my feed in last 2 days: "Kimi K3 beats Claude Fable 5 on this benchmark." So I trained a 0.8B model to beat GPT-4 on HumanEval. 98.8% vs 67%, on a single RTX 3060. Took 32 minutes. It's completely fake. This is why you need to be more careful on benchmark numbers 👇

True !!! NVidia milks the hype so well. And even if you somehow got 8 6000s, and somehow got around with bandwidth issues, you will now get choked on SM109 compatibility, and if you gor an old GPU say for reasonable price then you will be choked with Silicon architecture and attention compatibility. They really have everyone locked in steps 😜

@samarknowsit @ivanfioravanti Yeah, it's simply... big models still need hardware out of reach of consumers.

Unpopular opinion RTX 6000 is too much for a single developer/small team on one side and too little to run a powerful large model on the other. We need more VRAM. No? I don’t have one, just thinking about this.