Emerging AITrending on The Verge AIGoogle DeepMind releases Genie 4 world model with audio and real-time control
TechCrunch AI, . 15-second read.
The Radar
New models, research, tools and launches worth watching early, and what's coming up next.
Emerging AI
Emerging AITrending on The Verge AI
AI TechTrending on Hacker NewsCloudflare fixes KV cache leak in GLM-5.3 AI model
blog.cloudflare.com, . 30-second read.
Emerging AITrending on Hacker NewsBrown student maps invisible labor networks behind AI training
brown.edu, . 30-second read.
AI TechGoogle DeepMind's hurricane model outperforms traditional forecasting
Techmeme, . 20-second read.
AI CultureAI agents pitch investigative stories to editors at news outlets
Wired AI, . 15-second read.
Emerging AITrending on Hacker NewsDecade-old CNN outperforms latest LLMs at skin cancer diagnosis
lynchlab.net, . 30-second read.
AI TechTrending on Hacker NewsStudy: AI models generate harder-to-read text than humans, prioritizing density over clarity
arxiv.org, . 35-second read.
AI TechBlock releases voice model that runs locally on MacBook
TechCrunch AI, . 25-second read.
Live from GitHub, Hugging Face and Product Hunt
Just shipped
New AI launches, repos, models and papers. Updated 9 minutes ago, refreshed every 30 minutes.
- GHGitHub2 hours agoChannelCooperStart/capri-ai🌃 Capri — Edgerunners-inspired AI agent for your desktop. Works with Claude Code, Codex, Antigravity & more.★ 168 stars
- GHGitHub16 hours agopatch-gg/anymoveMotion capture for any creature from one AI video: two cameras in one frame, a DINOv2 joint reader, the rig fitted in both views, a GLB out.★ 176 stars
- PProduct Hunt17 hours agoOvermindOpen-source platform for continuous learningâ–² 270 upvotes
- PProduct Hunt17 hours agoOpenRadViewFree, open-source viewer for your own CT, MRI and X-raysâ–² 91 upvotes
- PProduct Hunt17 hours agoSpacesheepDocuments of the future, made by AI, read by your teamâ–² 181 upvotes
- PProduct Hunt17 hours agoPulseDevClick any UI bug on localhost and your AI agent fixes itâ–² 113 upvotes
- PProduct Hunt17 hours agoNuvio 5.0Hand your financial admin to your AI agentâ–² 113 upvotes
- PProduct Hunt17 hours agoTheySaid 4.0See how users will react, right where you build.â–² 316 upvotes
- vRelease24 hours agoollama v0.40.12[LFM3](https://ollama.com/library/lfm3) is Liquid AI's new family of small models built for local agents, tool calling, and structured output, on Liquid's new t
- vRelease1 day agotransformers v5.21MiniCPM-OCR is a document-understanding model from OpenBMB's MiniCPM-V family. It keeps the MiniCPM-V 4.6 architecture, which pairs a SigLIP vision encoder and
- HFNew paper2 days agoSkill^{infty}: Recursive Self-Improvement for Agent Skill OptimizationRecursive self-improvement (RSI), in which a system also improves the way it improves, is a long-standing aspiration on the path to AGI. Language agents increasâ–² 3 upvotes
- HFNew paper2 days agoEmbodiedLab: Benchmarking Embodied Agents That Act to KnowEmbodied agents are expected to accomplish real-world goals by deciding for themselves how and when to interact with their surroundings. For many such goals, châ–² 9 upvotes
- HFNew paper2 days agoThinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM AgentsLanguage model agents often abandon correct work when a user challenges it, yet simply telling them to stand firm makes them ignore real mistakes. Once an agentâ–² 7 upvotes
- HFNew paper2 days agoThinkBit: Search Codewords, Recalibrate Scales, Select by ReasoningReasoning language models generate thousands of tokens per answer, and generation speed is limited by memory bandwidth, so storing weights in two bits instead o
- HFNew paper2 days agoTrustworthy Recursive Self-Improvement in the Era of Large Models: Foundation, Risks, Evaluation and Open ChallengesAI systems built around large models increasingly use experience and feedback to make persistent changes to models, prompts, memories, tools, and workflows. Recâ–² 18 upvotes
- GHGitHub2 days agocodician-team/diffineryOpen-source, agent-first GitHub and GitHub Actions alternative that runs on your own Cloudflare account: git hosting, pull requests, code review and drop-in Git★ 54 stars
- GHGitHub3 days agoemanuele-em/kanzoSelf-hosted kanban for humans and agents. One 4 MB binary, starts in 2 ms, zero setup.★ 77 stars
- GHGitHub3 days agopatch-gg/trompeTexture a whole game level from a gray blockout with image models: plan cameras, paint, project, bake. Real shadows, glTF out.★ 171 stars
- HFTrending model4 days agoNanbeige/Nanbeige4.3-33Btext generation modelâ–² 665 likes
- vRelease4 days agoollama v0.40.11Ollama can now generate [speech](https://docs.ollama.com/capabilities/speech) on macOS. `ollama run` plays text or a chat model's reply aloud as the audio is ge
- HFTrending model5 days agoISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUFtext generation modelâ–² 391 likes
- HFTrending model6 days agoRecursiveSI/flinttext generation modelâ–² 969 likes
- HFTrending model9 days agoXHToken/Spark-X2.6-4Btext generation modelâ–² 781 likes
- HFTrending model11 days agoMiniMaxAI/MiniMax-M3.1-Flashimage text to text modelâ–² 413 likes
- GHGitHub2 hours agoChannelCooperStart/capri-ai🌃 Capri — Edgerunners-inspired AI agent for your desktop. Works with Claude Code, Codex, Antigravity & more.★ 168 stars
- GHGitHub16 hours agopatch-gg/anymoveMotion capture for any creature from one AI video: two cameras in one frame, a DINOv2 joint reader, the rig fitted in both views, a GLB out.★ 176 stars
- PProduct Hunt17 hours agoOvermindOpen-source platform for continuous learningâ–² 270 upvotes
- PProduct Hunt17 hours agoOpenRadViewFree, open-source viewer for your own CT, MRI and X-raysâ–² 91 upvotes
- PProduct Hunt17 hours agoSpacesheepDocuments of the future, made by AI, read by your teamâ–² 181 upvotes
- PProduct Hunt17 hours agoPulseDevClick any UI bug on localhost and your AI agent fixes itâ–² 113 upvotes
- PProduct Hunt17 hours agoNuvio 5.0Hand your financial admin to your AI agentâ–² 113 upvotes
- PProduct Hunt17 hours agoTheySaid 4.0See how users will react, right where you build.â–² 316 upvotes
- vRelease24 hours agoollama v0.40.12[LFM3](https://ollama.com/library/lfm3) is Liquid AI's new family of small models built for local agents, tool calling, and structured output, on Liquid's new t
- vRelease1 day agotransformers v5.21MiniCPM-OCR is a document-understanding model from OpenBMB's MiniCPM-V family. It keeps the MiniCPM-V 4.6 architecture, which pairs a SigLIP vision encoder and
- HFNew paper2 days agoSkill^{infty}: Recursive Self-Improvement for Agent Skill OptimizationRecursive self-improvement (RSI), in which a system also improves the way it improves, is a long-standing aspiration on the path to AGI. Language agents increasâ–² 3 upvotes
- HFNew paper2 days agoEmbodiedLab: Benchmarking Embodied Agents That Act to KnowEmbodied agents are expected to accomplish real-world goals by deciding for themselves how and when to interact with their surroundings. For many such goals, châ–² 9 upvotes
- HFNew paper2 days agoThinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM AgentsLanguage model agents often abandon correct work when a user challenges it, yet simply telling them to stand firm makes them ignore real mistakes. Once an agentâ–² 7 upvotes
- HFNew paper2 days agoThinkBit: Search Codewords, Recalibrate Scales, Select by ReasoningReasoning language models generate thousands of tokens per answer, and generation speed is limited by memory bandwidth, so storing weights in two bits instead o
- HFNew paper2 days agoTrustworthy Recursive Self-Improvement in the Era of Large Models: Foundation, Risks, Evaluation and Open ChallengesAI systems built around large models increasingly use experience and feedback to make persistent changes to models, prompts, memories, tools, and workflows. Recâ–² 18 upvotes
- GHGitHub2 days agocodician-team/diffineryOpen-source, agent-first GitHub and GitHub Actions alternative that runs on your own Cloudflare account: git hosting, pull requests, code review and drop-in Git★ 54 stars
- GHGitHub3 days agoemanuele-em/kanzoSelf-hosted kanban for humans and agents. One 4 MB binary, starts in 2 ms, zero setup.★ 77 stars
- GHGitHub3 days agopatch-gg/trompeTexture a whole game level from a gray blockout with image models: plan cameras, paint, project, bake. Real shadows, glTF out.★ 171 stars
- HFTrending model4 days agoNanbeige/Nanbeige4.3-33Btext generation modelâ–² 665 likes
- vRelease4 days agoollama v0.40.11Ollama can now generate [speech](https://docs.ollama.com/capabilities/speech) on macOS. `ollama run` plays text or a chat model's reply aloud as the audio is ge
- HFTrending model5 days agoISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUFtext generation modelâ–² 391 likes
- HFTrending model6 days agoRecursiveSI/flinttext generation modelâ–² 969 likes
- HFTrending model9 days agoXHToken/Spark-X2.6-4Btext generation modelâ–² 781 likes
- HFTrending model11 days agoMiniMaxAI/MiniMax-M3.1-Flashimage text to text modelâ–² 413 likes
- GHGitHub2 hours agoChannelCooperStart/capri-ai🌃 Capri — Edgerunners-inspired AI agent for your desktop. Works with Claude Code, Codex, Antigravity & more.★ 168 stars
- GHGitHub16 hours agopatch-gg/anymoveMotion capture for any creature from one AI video: two cameras in one frame, a DINOv2 joint reader, the rig fitted in both views, a GLB out.★ 176 stars
- PProduct Hunt17 hours agoOvermindOpen-source platform for continuous learningâ–² 270 upvotes
- PProduct Hunt17 hours agoOpenRadViewFree, open-source viewer for your own CT, MRI and X-raysâ–² 91 upvotes
- PProduct Hunt17 hours agoSpacesheepDocuments of the future, made by AI, read by your teamâ–² 181 upvotes
- PProduct Hunt17 hours agoPulseDevClick any UI bug on localhost and your AI agent fixes itâ–² 113 upvotes
- PProduct Hunt17 hours agoNuvio 5.0Hand your financial admin to your AI agentâ–² 113 upvotes
- PProduct Hunt17 hours agoTheySaid 4.0See how users will react, right where you build.â–² 316 upvotes
- vRelease24 hours agoollama v0.40.12[LFM3](https://ollama.com/library/lfm3) is Liquid AI's new family of small models built for local agents, tool calling, and structured output, on Liquid's new t
- vRelease1 day agotransformers v5.21MiniCPM-OCR is a document-understanding model from OpenBMB's MiniCPM-V family. It keeps the MiniCPM-V 4.6 architecture, which pairs a SigLIP vision encoder and
- HFNew paper2 days agoSkill^{infty}: Recursive Self-Improvement for Agent Skill OptimizationRecursive self-improvement (RSI), in which a system also improves the way it improves, is a long-standing aspiration on the path to AGI. Language agents increasâ–² 3 upvotes
- HFNew paper2 days agoEmbodiedLab: Benchmarking Embodied Agents That Act to KnowEmbodied agents are expected to accomplish real-world goals by deciding for themselves how and when to interact with their surroundings. For many such goals, châ–² 9 upvotes
- HFNew paper2 days agoThinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM AgentsLanguage model agents often abandon correct work when a user challenges it, yet simply telling them to stand firm makes them ignore real mistakes. Once an agentâ–² 7 upvotes
- HFNew paper2 days agoThinkBit: Search Codewords, Recalibrate Scales, Select by ReasoningReasoning language models generate thousands of tokens per answer, and generation speed is limited by memory bandwidth, so storing weights in two bits instead o
- HFNew paper2 days agoTrustworthy Recursive Self-Improvement in the Era of Large Models: Foundation, Risks, Evaluation and Open ChallengesAI systems built around large models increasingly use experience and feedback to make persistent changes to models, prompts, memories, tools, and workflows. Recâ–² 18 upvotes
- GHGitHub2 days agocodician-team/diffineryOpen-source, agent-first GitHub and GitHub Actions alternative that runs on your own Cloudflare account: git hosting, pull requests, code review and drop-in Git★ 54 stars
- GHGitHub3 days agoemanuele-em/kanzoSelf-hosted kanban for humans and agents. One 4 MB binary, starts in 2 ms, zero setup.★ 77 stars
- GHGitHub3 days agopatch-gg/trompeTexture a whole game level from a gray blockout with image models: plan cameras, paint, project, bake. Real shadows, glTF out.★ 171 stars
- HFTrending model4 days agoNanbeige/Nanbeige4.3-33Btext generation modelâ–² 665 likes
- vRelease4 days agoollama v0.40.11Ollama can now generate [speech](https://docs.ollama.com/capabilities/speech) on macOS. `ollama run` plays text or a chat model's reply aloud as the audio is ge
- HFTrending model5 days agoISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUFtext generation modelâ–² 391 likes
- HFTrending model6 days agoRecursiveSI/flinttext generation modelâ–² 969 likes
- HFTrending model9 days agoXHToken/Spark-X2.6-4Btext generation modelâ–² 781 likes
- HFTrending model11 days agoMiniMaxAI/MiniMax-M3.1-Flashimage text to text modelâ–² 413 likes
AI of the DayResearchTuesday, October 27
Claude's Formal Verification Work
AI-driven formal verification could reshape how software security gets checked
What it is
Anthropic used Claude to formally prove security properties across compiled code in 123 open-source libraries, uncovering bugs that had gone undetected for years.
Why now
Formal verification has historically been too labor-intensive for most software, but this shows LLMs can do it at a scale and speed that was previously impractical.
What's next
Watch whether this approach gets applied more broadly to critical infrastructure code, potentially closing the gap between what's verified today and what experts would ideally check.