What just shipped in AI
Trending now updated ✍️ Post about a trendPost it
Hacker NewsMeta takes down a critical video about meta AI Glasses after filming at MetaExplainedJev vs LLM Hacker NewsFeds Target AI Critics as "Foreign Agents" Hugging Facelaya by convaiinnovations GitHubZCode Daily PapersSpeakerMem-R1: Speaker-Centered Dual-Track Memory for Multi-Party Dialogue#1Gemini 3.8#2GPT-6 Astra#3Claude Opus 5.5#4GPT-6 Sol#5GPT-6 LunaExplained744B on a laptopNew modelEmber-1New modelGLM 5.3 PrimeNew modelQwen3.8 Max PrimeAgentsVoiceSecurityBenchmarksPeople & skills
Just inIn the newsWhite House tells OpenAI and Anthropic to let U.S. review new models before sharing them with British testersThe White House wants OpenAI and Anthropic to hold back new AI models from the U.K.'s AI Safety Institute until U.S. agencies get to review them first. The article White House tells OpenAI and Anthropic to let U.S.The Decodercoverage
In the newsLightspeed targets $250M for new India fund, focusing on early-stage AIThe venture firm is aligning its India fundraising cycle with its global funds for the first time, as it shifts to a shorter investment period.TechCrunch AIcoverageNew researchFraming by Wording, Framing by Selection: A Large-Scale Two-Dimensional Audit of French News Headlines, 2022-2025News headlines frame public issues both by what they select and by how they word it, yet computational framing work typically collapses these operations into a single score.arXiv cs.CLNew researchReward Hacking Challenges Oversight of Autonomous Research AgentsAutonomous research agents can design experiments, evaluate results, and write reports, giving them control over both a scientific result and the evidence used to support it.arXiv cs.CLNew researchStable and Faithful Explanations for Knowledge TracingKnowledge tracing (KT) models predict student performance opaquely, limiting pedagogical action. This study contributes a validation protocol testing predictive competitiveness (RQ1), explanation stability (RQ2) and…arXiv cs.LGNew researchSMILESGNN: Interpretable Clinical Toxicity Prediction via SMILES-Graph Cross-Attention FusionDrug toxicity prediction is critical for reducing late-stage attrition in drug discovery, yet remains challenging due to severe class imbalance, scaffold-based generalization, and the clinical need for interpretable predictions.arXiv cs.LGNew researchWhen Should Forecasting Agents Reason? Behavioral Stress Tests for Reliability RoutingForecasting agents increasingly combine language-model reasoning, retrieval, ensembling, and calibration, but it remains unclear when each behavior should be trusted.arXiv cs.AINew researchTW3Cast: A Frozen Router of Lightly Fine-Tuned Foundation Models for Time-Series Forecasting on GIFT-Eval, Selected Entirely on the Training SplitTW3Cast is a time-series forecasting system that reaches position 3 of 130 entries on the GIFT-Eval benchmark by mean MASE rank, as of 2026-09-14.arXiv cs.AINew releaseOllama v0.40.0 is outNew: Models run on MLX on Apple Silicon by default · In this release, model architectures supported by the MLX runner will run by default on Apple Silicon devices · ollama pull qwen3.8OllamaNew researchTraining Object Permanence in World ModelsObject permanence and solidity are hallmarks of human cognitive priors. Recent studies show that video generation models, a paradigmatic class of current world models, have begun to show emerged reasoning abilities, making them…Hugging Face Daily PaperscoverageNew releaseColibrì 1.12.1 is outNew: 96 pull requests since v1.12.0, 80 of them from contributors. Two tokenizers · brought back to the reference, brio on the ninth engine, coli chat working · again at the default context on two families, and a placement…Colibrì
In the newsTop AI experts badly underestimated how fast the field is moving, study findsLeading AI experts have consistently underestimated how fast AI is advancing, according to the Forecasting Research Institute. AI reached gold-medal level at the International Mathematical Olympiad five years ahead of the median…The Decodercoverage
In the newsBring your co-founder, partner, or colleague and get 50% off a second TechCrunch Disrupt 2026 passBuy one pass to TechCrunch Disrupt 2026 and get 50% off a second of the same ticket type. Register before event starts on October 13 at 8 a.m. PT.TechCrunch AIcoverage
In the newsPrismML brings its tiny LLMs to Qualcomm-powered smart glassesPrism's larger goal is open-weight AI that runs on devices and makes better use of the computing power they already have.TechCrunch AIcoverageAWSAWS Lambda durable functions are now available in AWS European Sovereign Cloud regionAWS Lambda durable functions are now available in the AWS European Sovereign Cloud. Lambda durable functions enable developers to build reliable multi-step applications and AI workflows within the Lambda developer experience.AWS What's New
In the newsOracle sends force majeure notice on its New Mexico Stargate data centerThe notice would allow Oracle to delay payments should the facility miss its 2028 target to come online.TechCrunch AIcoverage
In the newsSakana AI hires Jürgen Schmidhuber, inventor of deep learning, world models, and your next ChatGPT updateTokyo-based Sakana AI has hired Jürgen Schmidhuber as Chief Scientific Advisor. Sakana calls him the "father of modern AI." He'll help lead the company's new RSI Lab, which works on recursive self-improvement, meaning AI that…The Decodercoverage
In the newsGoogle's Suncatcher project aims to put AI data centers in orbit powered by solar energyGoogle's "Suncatcher" project aims to run AI infrastructure in orbit on solar power. A fridge-sized experimental satellite is set to launch on a SpaceX Falcon 9 on October 1.The Decodercoverage
In the newsMeta’s Muse Charm looks like a Tamagotchi, but it’s tapping into a much newer trendMeta’s new AI gadget may look like a Tamagotchi, but its dangling form factor taps into a much broader Gen Z trend around bag charms, retro tech, and turning gadgets into fashion accessories.TechCrunch AIcoverage
In the newsBlack Forest Labs launches FLUX 3 Action, an open robotics AI modelBlack Forest Labs is entering robotics with FLUX 3 Action. The open-world-action model uses camera feeds to predict what action a robot should take next.The Decodercoverage
In the newsGoogle Photos ‘Clueless’-inspired virtual closet is now available on Android and iOSThe AI-powered feature builds a virtual wardrobe from your photos, and is now broadly available after first rolling out to Android users in June.TechCrunch AIcoverage
Google DeepMindIntroducing Gemini 3.8 Live with Live AvatarGoogle DeepMind
AWSSpeaker-labeled transcription with WhisperX on SageMaker AIAny team working with spoken audio hits the same wall with generic speech-to-text. Think contact-center calls, all-hands meetings, podcasts, depositions, and broadcast media.AWS Machine Learning blog
AWSBuild a multi-account AI agent with AgentCore Gateway and MCPEnterprises increasingly want AI agents that can reason over data spread across many AWS accounts without copying or centralizing it.AWS Machine Learning blog
AWSAderant builds intelligent ticket triage with Amazon NovaThis guest post is co-written by Angela Mapes and Adam Walker of Aderant. In this post, we share how Aderant , a global provider of business management software for the legal industry, built an intelligent ticket triage system…AWS Machine Learning blog
NVIDIAEfficient MoE Training for Biological Foundation ModelsAs language models grow, scaling dense architectures becomes increasingly expensive. In a dense transformer, every token passes through every layer, so addingNVIDIA Developer
Google CloudPower your agents: Gemini 3.8 Live with Live Avatar is now generally availableFollowing our announcement of Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking last week, we are thrilled to share that Gemini 3.8 Live with Live Avatar is now generally available in Gemini Enterprise .Google Cloud AI & ML
Google CloudHow growing Latin American midsize businesses are building in the AI eraLatin America’s small and medium-sized businesses are the heartbeat of the region's economy — accounting for more than 60% of total employment in the region, according to United Nations estimates.Google Cloud AI & ML
Hugging FaceAccelerating vision-language models with LFM2.5-VL-DSparkHugging Face
Google CloudThe three things today's hottest startups are looking for in their AI stackGoogle Cloud has become the platform of choice for startups building AI. Our uniquely complete stack — including a choice of first- and third-party compute and models ; our platform for building and managing agents; and our…Google Cloud AI & MLHacker News buzzMeta takes down a critical video about meta AI Glasses after filming at Meta611 points and 367 comments on Hacker News.Hacker NewscoverageAWSAmazon SageMaker HyperPod Inference Gateway for scalable LLM inferenceAmazon SageMaker HyperPod Inference Gateway is a Kubernetes-native, GPU-aware routing system that deploys as a single EKS managed add-on on existing SageMaker HyperPod infrastructure with zero application changes.AWS What's NewAWSRun interactive workloads on Amazon EMR on EKS with Spark ConnectAmazon EMR on EKS now supports interactive Apache Spark sessions with Spark Connect. Data engineers and data scientists can develop and debug Apache Spark applications interactively from managed notebooks in Amazon SageMaker…AWS What's NewNew releaseOllama v0.34.4 is outStructured outputs on thinking models now apply in a single pass, making them faster and more reliable.Ollama
Hacker News buzzFeds Target AI Critics as "Foreign Agents"380 points and 423 comments on Hacker News.Hacker NewscoverageNew modelEmber-1 is now available$3 in · $15 out per 1M tokens. 1M context. Ember-1 is a specialized reasoning model from Fireworks Research, built on [Kimi K3](https://openrouter.ai/moonshotai/kimi-k3).FireworksNew researchSpeakerMem-R1: Speaker-Centered Dual-Track Memory for Multi-Party DialogueLong-term conversational memory in multi-party settings requires more than retrieving relevant content from long-term conversations: it must distinguish who said what, whom each statement concerns, how individuals perceive one…Hugging Face Daily PaperscoverageNew researchSpatial-Interactor: Learning Spatial Reasoning through Interaction with the Observable Physical WorldSpatial reasoning is essential for vision-language models (VLMs) to understand and act in the physical world. Reasoning in dynamic environments requires VLMs to perceive local state transitions caused by object motion and…Hugging Face Daily PaperscoverageNew researchHappyWorld-BenchEvaluating world models requires assessing both the quality of the worlds they generate and their consistency and responsiveness under exploration, interaction, and modification.Hugging Face Daily PaperscoverageNew researchThe Past Frames the Future: Memory for Autoregressive Video GenerationAdvances in generative models have improved video fidelity, enabling long-horizon generation, interactive world modeling, and evolving visual environments.Hugging Face Daily PaperscoverageNew releaseGemini CLI v0.61.0 is outNew: Changelog for v0.60.0-preview.0 · Changelog for v0.59.0 · preserve explicit versioned Flash model IDsGemini CLI
NVIDIAIntroducing NV-Reason-CT Open 3D CT VLM for Radiologist Chain-of-Thought ReasoningRadiology AI has made remarkable strides in detecting abnormalities across chest X-rays, pathology slides, and 2D scans. Yet one of the most clinically rich andNVIDIA DeveloperNew modelGLM 5.3 Prime is now available$2.8 in · $8.8 out per 1M tokens. 1M context. GLM-5.3-Prime is the high-speed variant of Z.ai's GLM-5.3, inheriting its full capabilities while delivering 1.5–2× the output throughput through inference acceleration.Z.aiNew releaseClaude Agent SDK v0.2.159 is outNew: Internal/Other Changes · Updated bundled Claude CLI to version 2.1.281 · Pinned default model for e2e tests to claude-opus-5 to work around CI failures with the CLI's new default modelClaude Agent SDK
NVIDIAValidate GPU Cluster Readiness Before AI Workloads LandA GPU cluster can pass every health check and still fail to run an AI workload. Even when every GPU, network link, and pod reports healthy, a 512-GPU trainingNVIDIA DeveloperNew modelQwen3.8 Max Prime is now available$4 in · $12 out per 1M tokens. 1M context. Qwen3.8 Max Prime is a higher-throughput variant of Qwen3.8 Max from Alibaba's Qwen team, served as a separate SKU at a higher price point. It accepts text, image, and video...Qwen
Hugging FaceHow to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning WorkflowsHugging Face
AWSFrom portal-hopping to instant answers: HEMA’s journey with MCP and Amazon BedrockThis post is co-written with Mauro Rallo and Patrick van der Plas from HEMA. When engineers at HEMA needed an answer, they went portal-hopping, navigating disconnected wikis, service catalogs, and IT portals to find it.AWS Machine Learning blog
NVIDIAManage Kubernetes Node Fleets with NodeWrightKubernetes manages what runs on your nodes. Managing the nodes themselves is the challenge: kernel settings, system packages, storage layouts, security agents,NVIDIA Developer
AWSAgentic conversational video intelligence built on AWSWith video intelligence powered by agentic AI, you can ask natural language questions about uploaded videos and get answers within seconds.AWS Machine Learning blog
AWSUse open weight models as your AI coding agent with Amazon BedrockAI coding agents have become a core part of how developers write, debug, and refactor software. Open weight models on Amazon Bedrock now make these agents practical to run privately and cost-effectively.AWS Machine Learning blogHacker News buzzClaude discovers a novel enzyme system with CRISPR-like repeats766 points and 786 comments on Hacker News.Hacker NewscoverageNew releaseLangGraph 0.4.32 is outNew: Changes since cli==0.4.31 · place self-hosted deployments on a listener · clarify agent flags and support env defaultsLangGraphGoogleGoogle Beam expands with new regions, partners, and customersWe’re expanding Google Beam to five new countries, and partnering with Industrious for an extended network.Google AI blogAWSAmazon Bedrock Managed Knowledge Base now supports Salesforce and Zendesk as native data source connectorsAWS announces Salesforce and Zendesk data source connectors for Amazon Bedrock Managed Knowledge Base, a fully managed retrieval-augmented generation (RAG) service.AWS What's New
In the newsGemini 3.8 TTS PlaygroundTool: Gemini 3.8 TTS Playground Google released two new Gemini text-to-speech models today - gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts .Simon Willisoncoverage
MicrosoftOffloaded inference for real-world physical AI roboticsAt a glance Challenges a core assumption in robotics AI: Our research shows that running physical AI inference exclusively on onboard GPUs can limit robot performance, battery life, and scalability, and that offloading inference…Microsoft Research
Google DeepMindAdvancing Private AI Compute with secure, server-side memoryIntroducing private, server-side memory to Private AI Compute for personal AI.Google DeepMindOpenAITwo years of OpenAI AcademyMarking two years of OpenAI Academy and bringing AI skills to even more communities.OpenAI
NVIDIAHow SWE-Serve Exposes the Gap Between Local Tests and Live ServingAn AI coding agent’s patch can pass tests yet fail when the server loads a real model and handles requests. Evaluating changes to inference-serving softwareNVIDIA Developer
MicrosoftYour architecture diagram is not your resilienceThis first article in our resilience series draws on a conversation with Mark Russinovich about how resilience is changing in the AI era and what it takes to continuously validate it at scale.Microsoft Azure blogHacker News buzzGemini 3.8 text-to-speech329 points and 148 comments on Hacker News.Hacker NewscoverageGoogle DeepMindGemini 3.8 text-to-speech says helloGoogle DeepMindHacker News buzzGPT-6 Astra has gained the ability to drive a car308 points and 246 comments on Hacker News.Hacker Newscoverage
MicrosoftDesigning agent-first platforms: What changes when agents do the workFor decades, applications have been designed to wait. A user clicks, a request arrives, code runs, a response goes back. And we got very good at this.Microsoft Azure blogNew modelSpace Bunny Alpha is now availableFree to use. 1M context. Space Bunny Alpha is an anonymous large model with blazing-fast inference, strong coding capabilities and native multimodal input support.OpenRouterNew modelAion 3.5 Mini is now available$0.7 in · $1.4 out per 1M tokens. 262K context. Aion 3.5 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It is the smaller, lower-cost sibling of Aion 3.5 and uses...AionLabsNew modelAion 3.5 is now available$3 in · $6 out per 1M tokens. 262K context. Aion 3.5 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models.AionLabsNew releasegolive-skill is rising on GitHub: 918 stars in 2 daysTake your agent-built product live: hosting, database, domain, email, payments — on your own accounts. Open-source Agent Skill + zero-dependency Node CLI: detect → plan → approve → apply → verify. No GoLive account, backend or telemetry. By mikehasa, in TypeScript.GitHub risingcoverageOpenAIOpenAI extends cyber access to Ukraine for civilian defenseOpenAI is extending access to its Daybreak program to the Government of Ukraine to support the cyber defense of civilian infrastructure.OpenAIHacker News buzzClaude Code reads AGENTS.md only when telemetry is on [fixed]480 points and 282 comments on Hacker News.Hacker NewscoverageOpenAISam Altman’s remarks at the United Nations Security CouncilOpenAI CEO Sam Altman discusses AI safety, human control, and international cooperation in remarks to the United Nations Security Council.OpenAIOpenAIHarvey turns legal context into stronger drafts with GPT-6 AstraGPT-6 Astra produces more structured, context-aware legal documents, freeing lawyers to focus on strategy.OpenAIOpenAIHow invideo improves color grading 3x with GPT‑6 AstraWith GPT‑6 Astra, invideo plans edits with greater precision, improves color correction and grading threefold, and produces 50 custom effects in one day.OpenAIOpenAIRingg’s AI agents resolve up to 65% of customer calls with OpenAIUsing GPT-5.6, Ringg powers multilingual agents across voice, chat, WhatsApp, and web for 90% less cost vs. GPT-4.1.OpenAINew modelSolar Mini 4 is now available$0.05 in · $0.2 out per 1M tokens. 524K context. Solar Mini 4 is Upstage's compact, cost-efficient language model, a 35B-parameter mixture-of-experts with 3B active parameters and a 524K context window.UpstageOpenAIIntroducing MentalHealthBenchMentalHealthBench is an expert-informed benchmark for evaluating helpful and safe AI responses across realistic mental health conversations.OpenAIIn the newsSF October 14th: A Birds of a Feather Session on Agentic EngineeringSF October 14th: A Birds of a Feather Session on Agentic Engineering I'm hosting an evening event with Jesse Vincent in San Francisco on Wednesday 14th October for people who are building weird and interesting things with and on…Simon WillisoncoverageOpenAIChatGPT Ads expands to Southeast Asia and TaiwanChatGPT Ads is expanding to Southeast Asia and Taiwan, giving eligible businesses new ways to reach people across more than 60 countries.OpenAINew releaseClaude Agent SDK v0.2.158 is outverbatim_prompts option : Added ClaudeAgentOptions.verbatim_prompts (default False ). When True , user messages are delivered to the CLI exactly as written — no @path file expansion and no slash-command dispatch.Claude Agent SDKAWSAmazon CloudWatch Omni: AI-first observability for agents and applicationsAWS announces the general availability of Amazon CloudWatch Omni, an evolution of Amazon CloudWatch. Omni is an AI-powered observability experience organized around your teams and the applications they run, so that you can…AWS What's NewVercel AI GatewayGemini 3.8 text-to-speech models now available on AI GatewayGemini 3.8 Flash-Lite TTS and Gemini 3.8 Flash TTS from Google are now available on AI Gateway . Both models take text and generate speech in more than 100 languages.Vercel AI Gateway
In the newsClaude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price warYesterday was Grok 4.7 ( pelicans ) and MiMo v2.6 Flash/Pro ( more pelicans ). Today Anthropic released Claude Opus 5.5 , and around an hour later OpenAI released GPT-6 Sol and GPT-6 Luna .Simon WillisoncoverageNew modelCommand A+ is now available$0.3 in · $1.5 out per 1M tokens. 192K context. Command A+ is Cohere's flagship model for enterprise agentic workflows. It accepts text and image inputs with a 192K context window, supports native tool calling with strict tool…Cohere
MicrosoftGPT-6 Astra, Sol, and Luna: For production agents in Microsoft FoundryToday, we are expanding our GPT-6 series by welcoming GPT-6 Sol and GPT-6 Luna to our generally available lineup in Microsoft Foundry .Microsoft Azure blogNew modelGPT-6 Luna Pro is now available$0.1 in · $0.5 out per 1M tokens. 1.1M context. GPT-6 Luna Pro is the same underlying model as [GPT-6 Luna](https://openrouter.ai/openai/gpt-6-luna), served with `reasoning.mode` set to `pro` for higher-quality responses on…OpenAINew modelGPT-6 Luna is now available$0.1 in · $0.5 out per 1M tokens. 1.1M context. GPT-6 Luna is the fast, cost-efficient model in OpenAI's GPT-6 series, positioned below GPT-6 Sol.OpenAINew modelGPT-6 Sol Pro is now available$2 in · $10 out per 1M tokens. 1.1M context. GPT-6 Sol Pro is the same underlying model as [GPT-6 Sol](https://openrouter.ai/openai/gpt-6-sol), served with `reasoning.mode` set to `pro` for higher-quality responses on complex…OpenAINew modelGPT-6 Sol is now available$2 in · $10 out per 1M tokens. 1.1M context. GPT-6 Sol is the cost-efficient high-end model in OpenAI's GPT-6 series, positioned below the flagship GPT-6 Astra and above the fast GPT-6 Luna tier.OpenAIAWSOpenAI GPT-6 Sol and GPT-6 Luna are now generally available on Amazon BedrockToday, AWS announces the general availability of GPT-6 Sol and GPT-6 Luna from OpenAI on Amazon Bedrock. Expanding the GPT-6 family alongside Astra, these two models give teams more ways to balance intelligence, speed, and cost…AWS What's New
NVIDIAEnabling Private High-Performance Production AI Inference with NVIDIA Confidential ComputingAs large language model (LLM) inference increasingly processes sensitive information and proprietary model context across personal, enterprise, and regulatedNVIDIA DeveloperMicrosoftClaude Opus 5.5 comes to Microsoft Foundry for long-running coding and knowledge workAI models are increasingly taking on work that extends far beyond a single prompt: building a feature across a codebase, investigating a complex issue, synthesizing hundreds of pages of information, or working through a…Microsoft Azure blogNew modelClaude Opus 5.5 is now available$4 in · $20 out per 1M tokens. 1M context. Claude Opus 5.5 is Anthropic's flagship model for demanding reasoning, coding, and long-horizon agentic work, succeeding Claude Opus 5.AnthropicNew releaseAnthropic Python SDK v1.8.0 is outapi: add support for claude-opus-5-5, inline tool definitions and MCP tool-list pinning (beta)Anthropic Python SDKAWSClaude Opus 5.5 is now available on AWS GovCloud (US)AWS GovCloud (US) now offers Claude Opus 5.5, Anthropic’s most capable Opus model yet and, the first of the Claude 5.5 model family, a better collaborator that handles long-running coding and knowledge work, reporting back…AWS What's NewAWSClaude Opus 5.5 is now available on AWSAWS now offers Claude Opus 5.5, Anthropic’s most capable Opus model yet and, the first of the Claude 5.5 model family, a better collaborator that handles long-running coding and knowledge work, reporting back clearly on what it…AWS What's NewAWSAWS Security Hub AI Inventory adds Azure self-hosted instance supportAWS Security Hub AI Inventory now supports discovering and cataloging AI assets running on self-hosted instances in Microsoft Azure.AWS What's NewNew releasevLLM v0.30.0 is outThis release features 762 commits from 315 contributors (104 new)!vLLMAWSAmazon Connect Customer launches agent-to-agent collaborationAmazon Connect Customer now supports agent-to-agent collaboration, giving customers the choice to bring in specialized AI agents during a live interaction to resolve a customer request.AWS What's NewHugging FaceHow UK AISI and EvalEval Are Making Benchmark Results ReproducibleHugging FaceHugging FaceTransformers now runs llama.cpp quantsHugging FaceHugging FaceJun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX communityHugging FaceVercel AI GatewayGPT-6 Sol and Luna now available on AI GatewayGPT-6 Sol and GPT-6 Luna from OpenAI are now available on AI Gateway . Both models bring GPT-6 improvements in professional work, coding, computer use, factuality, and communication at a lower price than GPT-6 Astra .Vercel AI GatewayVercel AI GatewayClaude Opus 5.5 now available on AI GatewayClaude Opus 5.5 from Anthropic is now available on AI Gateway . It is a step-change improvement over Opus 5 , with its biggest gains in agentic coding, long-running agent tasks, and knowledge work.Vercel AI GatewayNew releaseunreal-agent is rising on GitHub: 1.9K stars in 3 daysAsync-first agent harness. By unreallabsai, in Go.GitHub risingcoverageNew modelMiMo-V2.6-Pro-UltraSpeed is now available$4.35 in · $8.7 out per 1M tokens. 1M context. MiMo-V2.6-Pro-UltraSpeed is the fast speed edition of Xiaomi's flagship foundation model, MiMo-V2.6-Pro.Xiaomi
MicrosoftImproving synthesis prediction of small molecules at scale with RetroChimeraAt a glance We report on the recent publication of our retrosynthesis model RetroChimera in the journal Nature (opens in new tab) .Microsoft ResearchNew releaseLangGraph 1.2.12 is outNew: Changes since 1.2.11 · add response_schema to interrupt() · type undeclared v3 stream projectionsLangGraphHugging FacePruning LLMs Like a Physicist: Block Removal as an Ising Optimization ProblemHugging FaceVercel AI GatewayMiMo V2.6 models now available on AI GatewayMiMo V2.6 Pro , MiMo V2.6 Flash , and MiMo V2.6 Pro UltraSpeed from Xiaomi are now available on AI Gateway . MiMo V2.6 combines coding, reasoning, and tool use with native text, image, audio, and video understanding.Vercel AI GatewayNew releaseColibrì 1.12.0 is outNew: colibri 1.12.0 · 81 pull requests since v1.11.0. A new way to ask a model a closed question, · a redesigned dashboard and landing page, and a long run of small failuresColibrìNew releaseHemmingway-1 by Altworld is trending on Hugging FaceLanguage model from Altworld, free to download. 646 likes and 5.0K downloads so far.Hugging Face trendingcoverageNew releaseZCode is rising on GitHub: 6.7K stars in 5 daysZ.ai's coding agent harness. Powerful, intelligent, extensible. By zai-org, in TypeScript.GitHub risingcoverageNew releaselaya-mlx is rising on GitHub: 6.3K stars in 6 daysNative MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API. By mizorewww, in Python.GitHub risingcoverageNew releaseai-engineering-interview-questions-company-wise is rising on GitHub: 808 stars in 6 daysYour Cheat Sheet For AI Engineering Interviews at Top AI Companies - Questions and Answers. By pallavi-shekhar, in Markdown.GitHub risingcoverageNew releaseAnthropic Python SDK v1.7.0 is outNew: api: add group with display_name to rate limits, deprecate group_type · tools: add compact_before_next_turn() to the tool runner · bedrock: raise an API error for eventstream exception and error framesAnthropic Python SDK
Google CloudChanging the game: Using agentic AI to secure infrastructure codeAI is accelerating software development at an unprecedented pace. But as code generation scales, so do the challenges of securing the code, especially emerging AI-based vulnerability exploitations.Google Cloud AI & MLGoogleNew experts join Google’s AI & Economy teamWe are expanding our AI & Economy team with world-class academic advisors, fellows, and core internal researchers.Google AI blogGoogleCo-creating the future of fashion with GoogleGoogle worked side-by-side with designers Jane Wade and Sergio Hudson to custom-design Google Flow tools to prep for NYFW.Google AI blogNew releaselaya by convaiinnovations is trending on Hugging FaceClassifier from convaiinnovations, free to download. 3.5K likes so far.Hugging Face trendingcoverageNew releaseOpenAI Agents SDK v0.22.3 is outNew: align conditional approvals with validated tool arguments · deliver tool-not-found output on server-managed resume · keep command paths POSIX on a Windows hostOpenAI Agents SDKGoogleMaking global data easier to exploreGoogle and the UN system have launched the UN System Data Commons, a new open platform making global statistics accessible and easy to search.Google AI blogNew releasevLLM v0.3.0 is outNew: Release vllm-proto 0.3.0vLLMNew releaseCrewAI 1.15.22 is outNew: Support aliases as connection identifiers · Record reasons for deployment creation failures · Collect human feedback and pause events in tracingCrewAIGoogle CloudHow Orange built FinOps accountability, and why agents are nextAt Orange , the leading France-based multinational telecom provider, there are days when engineering teams set aside their delivery backlogs and spend the day cleaning up cloud spend together. There's a leaderboard.Google Cloud AI & MLGoogle CloudCloud CISO Perspectives: How Google monitors AI threats and advances AI defensesWelcome to the first Cloud CISO Perspectives for September 2026. Today, Sandra Joyce shares the latest details on Google’s visibility into how attackers are using AI, and how we’re using AI to stop them.Google Cloud AI & MLMistralMistral and Mozilla are bringing open, private and multilingual AI to your web browserOpen, private and multilingual AI is coming to your web browser. Mistral and Mozilla team up to put powerful, trustworthy AI where you already browse.Mistral AINew releaseXing4.0-29B-A4B by XingChen-AGI is trending on Hugging FaceLanguage model from XingChen-AGI, free to download. 1.6K likes and 43K downloads so far.Hugging Face trendingcoverageNew releaseQwen-Image-2.1 by Comfy-Org is trending on Hugging FaceOpen model from Comfy-Org, free to download. 701 likes and 3266K downloads so far.Hugging Face trendingcoverageGoogle DeepMindIntroducing Gemini 3.8 Live and 3.8 Live Extended ThinkingGoogle DeepMind
GoogleAI for Societal ImpactExplore this collection to see how experts and local leaders are using AI breakthroughs to ensure everyone can share the opportunity of AI.Google AI blogNew releaseQwen-Image-2.1 by Qwen is trending on Hugging FaceImage generator from Qwen, free to download. 2.2K likes and 42K downloads so far.Hugging Face trendingcoverage
MicrosoftThe Economics of Agent Optimization: How AI agent governance controls cost and proves ROIThis blog post is the fourth and final installment of The Economics of Agent Optimization , which shares the strategies, capabilities, and proof points that can help you optimize agent costs and run AI as a managed investment…Microsoft Azure blogNew releaseTransformers 5.17.0 is outNew: Release v5.17.0 · New Model additions · Hy4-Preview is a 780B-parameter mixture-of-experts language model that activates 49B parameters perTransformersMistralCloudera and Mistral Partner to Bring Specialized, Sovereign Intelligence to Enterprise DataCloudera and Mistral join forces to bring specialized, sovereign AI intelligence to enterprise data, helping regulated industries innovate on their own terms.Mistral AINew releaseCrewAI 1.15.21 is outNew: Add telemetry to track checkpoint runtime and CLI usage · Fix gateway errors reported inside an HTTP 200 response · Keep deploy push on the AMP create sourceCrewAINew releaseOpenAI Agents SDK v0.22.2 is outNew: support current image generation tool options · prevent UnixLocal file API symlink races · reset compaction response chain after popOpenAI Agents SDKMistralModernizing complex legacy code with AI agents.Mistral helped a European energy operator migrate 40,000 lines of Fortran 77 to C++. Learn how it was done, and the lessons to carry forward.Mistral AI
Google DeepMindAlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genomeAlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.Google DeepMindMistralMistral raises €3B to make sovereign, open-weight AI the technology frontierMistral today announced that it has raised €3 billion in a Series D funding round at a post-money valuation of more than €21 billion.Mistral AINew releaseZDTaichu5.0-9B by TaichuAI is trending on Hugging FaceVision-language model from TaichuAI, free to download. 1.0K likes and 9.5K downloads so far.Hugging Face trendingcoverage
MicrosoftGigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation modelsAt a glance The Flash family extends GigaPath and GigaTIME with dramatically improved efficiency, making large-scale pathology research more accessible and practical.Microsoft ResearchNew releaseTransformers v5.16.0 is outNew: Release v5.16.0 · New Model additions · Qwen4-Exp builds on Qwen3.5's hybrid text and multimodal architecture with three key components: GatedResidual (GR), Qwen Sparse Attention (QSA), and Per-Layer Embedding (PLE)Transformers
MicrosoftMindTopo reveals VLMs’ spatial reasoning abilitiesAt a glance MindTopo is a new benchmark for testing topological reasoning in AI, evaluating whether multimodal models can understand concepts such as connectivity, enclosure, order, separation, and knots.Microsoft ResearchNew releaseModel Context Protocol 2026-07-28 is outThis release marks the stable release of the 2026-07-28 revision of the Model Context Protocol.Model Context ProtocolNew releaseModel Context Protocol 2025-11-25 is outThis release marks the stable release of the 2025-11-25 revision of the Model Context Protocol.Model Context Protocol