Great application of small language models from Salesforce and colleagues. AI-scientist pipelines are well known to struggle to propose novel and diverse research ideas. This team trains an 8B reasoning model to write several research ideas in one generation from a set of source papers. The reward scores the whole set, weighting each idea's quality by how far it sits from the ideas written before it. The model learns to move each new idea away from the earlier ones. On their contamination-free benchmark, the 8B model scores higher on creative utility, a measure that combines quality and diversity, than Claude Opus 4.8 and GPT-6 Astra. arxiv.org/abs/2610.22324 Chat with Paper: academy.dair.ai/papers/trainin…
✈ DeepSeek-V4.1-Flash-1017 is on Ollama! US-based hosting with zero data retention. It's strong on agentic tasks while being cheaper per task. Try it on Ollama Cloud: ollama run deepseek-v4.1-flash:1017-cloud Claude Code: ollama launch claude --model deepseek-v4.1-flash:1017-cloud OpenCode: ollama launch opencode --model deepseek-v4.1-flash:1017-cloud DeepSeek @deepseek_ai 🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness , Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4 🔗 View Quoted Tweet
Ben Horowitz on what it would take to actually see an AI "fast takeoff" scenario, using a framework from Databricks CEO @alighodsi : 1. Each new model needs far fewer GPUs than the last 2. Each new model takes less time to train 3. Each new model is more accurate than the last 4. It keeps happening, again and again "Zero of those four have been met... We're really talking about a risk that's entirely theoretical at this point." @bhorowitz on @theallinpod 🔗 View on Twitter
OpenCode 遭入侵，1.4% 的 Console 用户受影响；没收到邮件就没受影响 OpenCode 10 月 17 日发帖说，前一天晚上发生了一起安全事件，和一次大规模“蒸馏攻击”有关，受影响的是 1.4% 的 OpenCode Console 用户。 怎么知道自己有没有受影响？看邮箱。官方说 17 日上午已经给所有受影响的用户发了邮件，没收到邮件就是没受影响。受影响的是 Console 里绑定了外部服务商的一部分用户，没绑定的不在其中。 收到邮件该怎么办，官方公开的帖子没写。据 OfficeChai、Notebookcheck 等科技媒体转述，邮件里说“工作区里保存的第三方服务商密钥（比如 OpenAI、Anthropic）可能已经泄露”，要求用户到各家服务商后台作废旧密钥、换成新的，更新用到这些密钥的程序，再查一查最近的用量有没有异常。这部分来自媒体报道，OpenCode 的公开帖里没有。 OpenCode 说事件还在处理中，暂时不能透露更多细节，正在和 Anthropic、OpenAI、Google 一起处理。 媒体报道： officechai.com/ai/opencode-di… OpenCode @opencode We had a security breach last night related to a large scale distillation attack that impacted 1.4% of OpenCode Console users We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked The incident is ongoing so we cannot share more details at this point but we are working with Anthropic, OpenAI, and Google to remediate 🔗 View Quoted Tweet
Working on getting our US hosts up and running. Very soon! DeepSeek @deepseek_ai 🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness , Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4 🔗 View Quoted Tweet
Perplexity Decider Flash is pretty good and can run super fast and on-device on your M4 Mac 16GB Wolfram Ravenwolf @WolframRvnwlf Perplexity Flash takes the crown! ⚡ In our tests, it matched Wald's accuracy at nearly twice the speed on a 16 GB Mac mini. Synthetic agent decisions, real email threads, prompt injections: see why Flash is our new pick for email-completion checks. 👇 x.com/i/article/2111… N 🔗 View Quoted Tweet
Product Shot turns your kitchen-counter photo into a studio shot without leaving ChatGPT. Even corn gets the campaign treatment. Pika @pika_labs Creative work demands a visual environment. So we partnered with OpenAI to turn ChatGPT into one. Introducing the Pika X ChatGPT plugin extension. All of Pika’s creative tools are now directly in your chat and side panel, in a new interactive creation experience. More on what’s new 👇 🔗 View on Twitter 🔗 View Quoted Tweet
Mind-blowing to see these Pareto improvements in performance and cost. More than anything, it's crazy how quickly these improvements are landing across providers. This matters a lot to builders. DeepSeek @deepseek_ai 🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness , Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4 🔗 View Quoted Tweet
Some strong embeddings models have been released recently. Ours is still better. We benchmarked Embed 5 Fast against Google's EmbeddingGemma 2 across 163 datasets using RCP-nDCG @10 . Embed 5 Fast leads on over 95% of them.
Agents change. Your context shouldn’t. Nick Beaird • @Nickanbe @akothari across every agent / tool i've used @NotionHQ has been the one constant 🔗 View Quoted Tweet
Gavin Baker on the other side of Ben's bandwidth story: at the peak of the 2000 bubble, 97% of America's fiber sat dark, and today there are no dark GPUs. "I had, depending on how you look at it, the privilege and the misfortune of being a tech investor during the year 2000 bubble, which was really a telecom bubble." "The year 2000 bubble was defined by something called 'dark fiber.' Dark fiber was literally fiber that was laid down in the ground, and not lit up... At the peak of the bubble, 97% of the fiber that had been laid in America was dark." "Contrast that with today. There are no dark GPUs. All you have to do is read any technical paper, and one of the biggest problems in a training run is that GPUs are melting." @GavinSBaker 🔗 View on Twitter a16z @a16z Ben Horowitz on the lesson LoudCloud, his early cloud company, taught him about today's AI neoclouds, and why they now face the opposite problem: "It can end up being like a very dangerous arbitrage business, where you can get upside-down easily. And so if you look at LoudCloud, just as an example, here's a simple thing that happened to us." "In 1999, bandwidth was $1,500 a megabit. And you couldn't get it. It was very hard to get it. And so you had to get a long-term contract, three years. By 2001, it was $50 a megabit... And if you borrowed money to get it, you're dead." "That kind of thing is so out of the control of the entrepreneur." "I think the core dynamic of AI has been working for CoreWeave, though, in that there's a massive shortage, not an overage of supply... Token prices have gone down, but GPU prices have not." "And by the way, data center space is going up in price because even labor to build data centers is hard. The politics are making it hard to build data centers. So the whole supply chain price is going up and demand is not subsiding." "Those are very big tailwinds for CoreWeave, despite the fact that the debt markets, kind of independent of them, are tough." @bhorowitz w/ @chamath on @allinpod 🔗 View on Twitter 🔗 View Quoted Tweet
Product Ad gave this armchair a full commercial, voiceover and all, straight out of ChatGPT. 🔗 View on Twitter Pika @pika_labs Creative work demands a visual environment. So we partnered with OpenAI to turn ChatGPT into one. Introducing the Pika X ChatGPT plugin extension. All of Pika’s creative tools are now directly in your chat and side panel, in a new interactive creation experience. More on what’s new 👇 🔗 View on Twitter 🔗 View Quoted Tweet
AI is helping Americans shop, but 42% don't trust it to make a purchase yet Even in the most popular categories, only about a quarter would let AI choose and buy for them: - 27% for clothing - 26% for electronics - 25% for travel Charts of the Week: a16z.news/p/this-week-in… a16z @a16z More than 1 in 3 Americans bought something based on an AI recommendation in the past year 44% say AI helps them find better value for money 34% would pay more for a product if AI showed it was the best option for them Charts of the Week: a16z.news/p/this-week-in… 🔗 View Quoted Tweet
You bring the idea, ChatGPT brings Pika, and Camera Director turns it into a full cinematic short. Director's chair not required. 🔗 View on Twitter Pika @pika_labs Creative work demands a visual environment. So we partnered with OpenAI to turn ChatGPT into one. Introducing the Pika X ChatGPT plugin extension. All of Pika’s creative tools are now directly in your chat and side panel, in a new interactive creation experience. More on what’s new 👇 🔗 View on Twitter 🔗 View Quoted Tweet
Catch more fraud. Decline fewer legitimate payments. Freddy Lecue of @WellsFargo reports better fraud detection and fewer false positives than their production XGBoost model. His AI Engineer New York clip explains why transaction history matters. #AIEngineerNY 🔗 View on Twitter
Your workflows have changed with AI. How about your setup? Physical controls are back. Take the new Logitech MX Keypad, now directly integrated with the GitHub Copilot app. This experience allows you to run agentic workflows with control, map a repeated prompt, start or resume a session, and help you stay in flow. github.blog/news-insights/… 🔗 View on Twitter
Why is speaking of words "doing work" and "doing heavy lifting" so common in AI-generated replies? I assume this phase will pass, but how did it happen at all? Is there one model with this idiosyncrasy that they all use?
Does anyone else stir their tea or coffee before putting the milk in instead of after? (I do because the spoons are near the kettle, but the milk is in the fridge on the other side of the kitchen.)
*Over $11.5M in perks and deals btw Jay @ LegalOS @jaykhan YC W27 applications are now open What YC gives you: → $500K investment → YC partners in your corner, for life → Faster launches, growth and customer acquisition → Demo Day in front of thousands of investors → Over $500K in credits and 100+ perks → Alumni include Airbnb, Stripe, DoorDash & Coinbase Truly a life-changing opportunity, one application away 🔸 🔗 View Quoted Tweet
Ali Ghodsi of @Databricks on the four conditions that would have to hold for AI to truly improve itself: "The next model requires less resources, less GPUs to train... The second condition: the next model takes less time to train as well. Third, accuracy of the model, the intelligence is increasing. And fourth, we can do the former three again and again and again... All at the same time, not just any of them." "If you just mean that the software is writing itself, we're already there today. Like 90-some percent of the software in Databricks is written by AI. Does it matter if the last few percent is also written by AI? No." "If you're getting these four conditions, then you might get a speed-up where the next model, let's say, takes half amount of time and half the resources, and it is more intelligent... By the way, I don't even know if that necessarily leads you to superintelligence." "A lot of people, their definition is just, 'Hey, if I'm not even coding anymore and it's coding itself'... I think they're conflating, 'Hey, what's my value and is it scary for me?' Versus, 'Hey, that then means we'll get that superintelligence that in 2014 was hypothesized by Bostrom.'" @alighodsi @martin_casado @sarah_wang 🔗 View on Twitter
Okay, so it's just compressed air. Easier. But crappy specific impulse (less than 50 seconds). Even assuming that 50% of the mass of the car is the compressed air, that gives you less than 10s of useful thrust (with thrust-to-weight ratio larger than 1.0). Probably more like 5 seconds, actually. So the car must be an empty, lightweight shell containing big compressed air tanks. Cute demo, but totally impractical. But of course, you knew that.
Recommended paper integrating decision models like Jev into harnesses. DAIR.AI @dair_ai Must-read paper integrating decision models like Jev into harnesses. They distilled an open 4B decision model, mostly from Jev's answers on the agent's own states, that beats Jev inside that agent and is much cheaper to run. Their harness, JevAgentHarness, sends each routine choice to the decision model and uses its answer only when the confidence clears a threshold fitted on the agent's own trajectories. Below the threshold, the LLM decides as usual. With the fitted threshold, tool and argument cascades cut cost by 14.2% without changing the outcome of any of 500 tasks. A fixed threshold of 0.5 lost 21.5 points. They also train AgentDecision-4B on states the agent reaches after its own mistakes, with Jev's answers as labels. On tasks unlike its training data, the agent solved 91.5% with it and 82.0% with Jev. On familiar tasks, 4B models trained this way made agents 81.1 to 89.0% cheaper and up to 2.5 times faster while staying within 3 points of success. Paper: academy.dair.ai/papers/jevagen… 🔗 View Quoted Tweet
To clarify: Replit remains model agnostic. Our users can choose Anthropic models as well as many other supported models. What I meant at TC Disrupt is that OpenAI is now our preferred partner powering Agent based on performance, cost, and strength of our partnership. We believe in giving our users model choice while optimizing the default experience for the best performance at the best price. TechCrunch @TechCrunch Replit CEO and co-founder Amjad Masad made the comments during TechCrunch Disrupt 2026. spr.ly/6013BHM92F 🔗 View Quoted Tweet
🧠 Yann LeCun helped invent the convolutional neural network and shared the 2018 Turing Award with Hinton and Bengio. Now he joins us at AI Dev NYC to share his latest work at AMI Labs. 📍 NYC 🗓️ Nov 30 to Dec 1 ai-dev.deeplearning.ai tVBz Just. Keep. Building.
Building safer AI agents takes more than model safeguards. @perplexity_ai open sourced SPACE, the agent sandbox platform behind Computer, giving developers more control over where agents run code and what they can access. Cheers to more teams open-sourcing tools to build safer autonomous AI agents. Perplexity @perplexity_ai We're open sourcing SPACE, the sandbox platform that runs every task in Perplexity Computer. SPACE gives each agent task its own Firecracker microVM and controls credentials and outbound traffic from outside the VM. Available now under Apache 2.0: github.com/perplexityai/S… 🔗 View Quoted Tweet