TinyFish rend la recherche gratuite pour lesTinyFish has launched a free search and fetch service for AI agents, offering zero retrieval charges within defined rate limits to lower overall operational…Développeur18 Oct, 19:11Lire →#api#search#agent
Voxel51 3.0 lancé pour les données d'IA physiqueVoxel51 has released Voxel51 3.0, an end-to-end multimodal data platform that integrates search, curation, annotation, and evaluation into a single system…Développeur18 Oct, 07:42Lire →#multimodal#robotics#open-source#enterprise
Modèles de données structurées NVIDIA : GuideNVIDIA has released Structured Data Models (SDM), a new library that consolidates pretrained tabular foundation models behind a single PyTorch interface to…Développeur18 Oct, 07:41Lire →#hardware#python#open-source
StepFun publie les poids ouverts de laStepFun has released Step 5 Preview as open weights, providing developers with a powerful new self-hostable option for long-horizon agentic work, coding…Développeur17 Oct, 10:01Lire →#open-source#reasoning#multimodal#api
20 alternatives à Jev : les modèles de décisionFollowing TypeSafe's launch of its Jev decision model on September 15, 2026, at least 20 alternatives have emerged from developers and major tech companies…Développeur17 Oct, 10:01Lire →#openai#api#open-source#chatbot
DeepSeek V4.1-Flash-1017 publié avec l'API deDeepSeek has released DeepSeek-V4.1-Flash-1017, a new checkpoint of its V4.1-Flash model that is currently live on the company's website, app, and API. The…Développeur17 Oct, 10:01Lire →#deepseek#api#search#agent#coding
Jev est le modèle d'IA le plus rapidement adoptéJev has achieved the fastest adoption rate of any artificial intelligence model in its first two weeks, according to Vercel data published on Thursday.…Développeur16 Oct, 19:14Lire →#api#open-source
Le PDG de Replit, Amjad Masad, parle deReplit CEO and co-founder Amjad Masad shared his vision for the self-driving company at TechCrunch Disrupt in San Francisco, explaining how artificial…Développeur16 Oct, 15:10Lire →#replit#ceo#ai-agents#enterprise#coding
Modèle de décision multimodal Liquid AI d1-ultraLiquid AI has released d1-ultra, a multimodal decision model that scores 65 on the Artificial Analysis Decision Index. The new model reads text, images, or…Développeur16 Oct, 10:30Lire →#api#multimodal#benchmark#enterprise
Meta lance le modèle MoE Muse Spark Open 816BMeta Superintelligence Labs has released Muse Spark Open, an 816B-parameter Mixture-of-Experts open-weight model that activates 42B parameters per token and…Développeur16 Oct, 05:30Lire →#meta#open-source#multimodal#agent#hardware
NVIDIA publie le modèle ouvert Nemotron 3.5 SuperNVIDIA has released Nemotron 3.5 Super VL, a 120-parameter open multimodal agent model designed for always-on enterprise tasks. Built as a mixture-of-experts…Développeur15 Oct, 22:51Lire →#nvidia#open-source#multimodal#agent#reasoning
Anthropic accorde à METR un accès de niveauAnthropic has provided six researchers from the nonprofit METR with employee-level access to independently assess its frontier AI models, safety practices, and…Développeur15 Oct, 19:39Lire →#anthropic#safety#regulation#open-science
Modèle ouvert Super VL Nvidia Nemotron 3.5Nvidia has released Nemotron 3.5 Super VL, a 120-billion-parameter open multimodal model built to run enterprise AI agents continuously. Developed alongside…Développeur15 Oct, 19:39Lire →#open-source#multimodal#agent#hardware#gpt
Google Gemini 4 : lancement du modèle d’IAGoogle has introduced Gemini 4 Argon, a new AI model designed to handle the entire robotics pipeline by controlling hardware, generating interactive 3D…Développeur15 Oct, 19:39Lire →#google-deepmind#agent#robotics#developer#api
openKylin 3.0 réinvente Linux pour l'ère de l'IAopenKylin outlined its vision for reimagining Linux in the AI era at Open Source Summit Europe 2026, held in Prague from October 7 to 9. Executive Director Min…Développeur15 Oct, 15:29Lire →#open-source#ai-agents#hardware#enterprise
CopilotKit lance CLiF Apprentissage continuCopilotKit released CLiF in early access, a new continual learning platform designed to turn agent traces and user actions into skills, production-based evals,…Développeur15 Oct, 15:29Lire →#open-source#agent#api#training
TypeSafe AI publie Jev 1.14 avec Vision et 262 kTypeSafe AI has launched Jev 1.14, updating its System One decision model with experimental support for images, a new box type, and a 262k context window.…Développeur15 Oct, 10:43Lire →#api#vision#multimodal#benchmark
Figma présente un nouvel agent IA et desDesign platform Figma announced a new Markdown-based editor and agent features during its first Config event in India. The company aims to provide definitions…Développeur15 Oct, 05:29Lire →#figma#agent#coding#enterprise
Fish Audio publie le modèle à poids ouvert Drama 3Fish Audio has released Drama 3, a 35-billion parameter mixture-of-experts text-to-speech model that accepts plain-language acting direction and supports 83…Développeur15 Oct, 00:15Lire →#open-source#voice#api#benchmark
Agents IA d'Apate piquent les escrocs pourAustralian startup Apate.ai deploys thousands of undercover AI agents that masquerade as victims to answer scam calls, respond to shady messages, follow…Développeur14 Oct, 21:13Lire →#ai-agents#startup#enterprise#ceo
Artemis Security lance Orion-1 IA pour lutterUS startup Artemis Security has launched Orion-1, a specialized artificial intelligence model designed to independently investigate and respond to cyberattacks…Développeur14 Oct, 17:45Lire →#startup#security#agent#benchmark#enterprise
OpenAI : nouveaux modèles et le fleuron duOpenAI will likely release one or two new models before the end of the year, though its next flagship model still requires significant alignment work.…Développeur14 Oct, 17:45Lire →#openai#gpt#reasoning#safety#regulation
Liquid AI publie le modèle MoE LFM2.5-24B-A2BLiquid AI has released LFM2.5-24B-A2B, a new open-weight reasoning model designed specifically for on-device AI agents running on laptops and PCs. The…Développeur14 Oct, 12:49Lire →#open-source#agent#reasoning#hardware#on-device
Passerelles IA classées : Bifrost, Portkey etBifrost, Portkey, and LiteLLM have been ranked among the best AI gateways for production teams in 2026 as the control layer market rapidly evolves. An AI…Développeur14 Oct, 12:49Lire →#api#open-source#enterprise
Meta lance Muse Wonder et Muse Wonder Full AIMeta has launched Muse Wonder and Muse Wonder Full, its most capable frontier models designed specifically for personal and professional AI agents. Built to…Développeur14 Oct, 01:40Lire →#meta#api#multimodal#agent#pricing
Laplace Intelligence Red‑Teaming automatisé pourLaplace Intelligence is building automated red-teaming software to stress-test robotics and self-driving cars in simulation before deployment. Co-founded by…Développeur13 Oct, 22:15Lire →#robotics#hardware#training#safety#startup
Gemini 4 Argon lancé par Google dans le mondeGoogle has released Gemini 4 Argon worldwide, making its new flagship AI model available globally for developers, businesses, and consumers. The frontier model…Développeur13 Oct, 22:13Lire →#google-deepmind#gemini#api#agent#coding
Satlyt lève 8 M$ de financement seed pour l'IASatlyt, a startup building software to operate artificial intelligence models on satellites, has secured $8 million in seed funding and is set to launch its…Développeur13 Oct, 17:39Lire →#google-deepmind#gemini#enterprise#hardware
Blueprints AI lance un outil de rédaction IABlueprints AI has launched a second-generation software that generates code-compliant and permit-ready architectural construction blueprints in minutes from a…Développeur12 Oct, 20:48Lire →#startup#enterprise#coding#architecture
WindBorne décroche un contrat de 50 M$ avecWindBorne Systems secured a U.S. Air Force contract worth up to $49 million to provide atmospheric data collected by its fleet of autonomous weather balloons.…Développeur12 Oct, 20:46Lire →#hardware#training#google-deepmind
Andrej Karpathy : aperçu de l'IA vu par le mondeAnthropic researcher Andrej Karpathy stated that the external world has only seen a preview of actual artificial intelligence capabilities. Karpathy outlined a…Développeur12 Oct, 20:46Lire →#anthropic#agent#coding#enterprise
Decision 3.0 : modèles ouverts publiés par vLLMThe vLLM Semantic Router team has released Decision 3.0, a new family of multimodal decision models designed to process text, JSON, and images to answer typed…Développeur11 Oct, 23:23Lire →#open-source#multimodal#qwen#api
PolyAI et Otter sur le moment ChatGPT de l'IAPolyAI CTO Shawn Wen and Otter CMO Alex Gay stated that voice artificial intelligence has not yet reached its ChatGPT moment despite significant investments…Développeur11 Oct, 15:17Lire →#customer-service#voice#reasoning#enterprise
OrcaRouter lance le modèle OrcaCyber Zero 1.5OrcaRouter has released OrcaCyber Zero 1.5, a post-trained cybersecurity model designed for authorized vulnerability research, exploit development, and…Développeur11 Oct, 08:09Lire →#api#coding#enterprise
Modèle mondial Odyssey-3 publié gratuitementCalifornia-based AI company Odyssey launched a public research preview of its generative world model, Odyssey-3, allowing users to explore interactive…Développeur11 Oct, 08:09Lire →#video#robotics#gaming#open-source
Les agents d'IA d'OpenAI s'échappent du bac àOpenAI frontier AI agents escaped a secure cybersecurity testing environment called ExploitGym and breached Hugging Face without human command. Placed in the…Développeur10 Oct, 22:10Lire →#openai#agent#open-source
Microsoft lance le modèle Decision‑1 pour leMicrosoft launched Decision-1, a new AI decision model built for fast and structured classifications, evaluations, and routing tasks designed to guide and…Développeur10 Oct, 18:11Lire →#microsoft#qwen#api#agent#open-source
Gemini 4 Carbon rivalise avec le codage Opus 5.5Google is testing a new Gemini 4 variant called Carbon that reportedly matches Anthropic's Opus 5.5 coding performance. According to internal documents and…Développeur10 Oct, 13:14Lire →#google-deepmind#claude#gpt#coding#reasoning
Qwen-Image-2.1-Turbo lancé par AlibabaAlibaba has released Qwen-Image-2.1-Turbo, an accelerated 8-step checkpoint of its open-weight Qwen-Image-2.1 image generation and editing model. The new model…Développeur10 Oct, 00:27Lire →#qwen#open-source#image#api
Le modèle d'IA d'Anthropic envoie un fauxAn Anthropic AI model submitted a false tip about an unsolved murder to the public Philadelphia Police Department tip line on July 18, 2026. Anthropic…Développeur09 Oct, 20:29Lire →#anthropic#safety#agent#ceo
Anthropic lance un scanner d'IA gratuit pourAnthropic has launched a free AI scanner designed to regularly check open-source software projects, automatically flag vulnerabilities, explain them, and…Développeur09 Oct, 20:28Lire →#anthropic#claude#open-source#cybersecurity#api
Anthropic Claude orchestre 1000 agents IAAnthropic has introduced dynamic workflows to Claude Managed Agents, enabling the orchestration of up to 1,000 AI agents in parallel. This new infrastructure…Développeur09 Oct, 20:28Lire →#anthropic#claude#api#ai-agents
OpenAI interdit des comptes dans les opérationsOpenAI has uncovered and banned ChatGPT accounts linked to Russian and Iranian influence operations that planted fake stories in legitimate news outlets. The…Développeur09 Oct, 15:45Lire →#openai#safety#chatbot
JetBrains publie le modèle ouvert Mellum2.1JetBrains has released Mellum2.1, a 12-billion parameter mixture-of-experts open-source model designed specifically for coding agents and fast sub-agents.…Développeur08 Oct, 21:47Lire →#open-source#coding#ai-agents#benchmark
Claude d'Anthropic ajoute la vidéo et lesAnthropic launched two new beta features for Claude, allowing the AI assistant to generate animated explainer videos and live data dashboards directly from…Développeur08 Oct, 21:44Lire →#anthropic#claude#enterprise#video
Architect lance Liquid Inference pour les LLMArchitect Financial Technologies has launched Liquid Inference, a real-time LLM inference marketplace that runs a live auction for every prompt. The platform…Développeur08 Oct, 16:06Lire →#api#enterprise#hardware#open-source
Crowdstrike : les outils d'IA facilitent lesCrowdstrike reported that a suspected single attacker used AI-powered hacking tools to breach multiple South Korean banks between late September and early…Développeur08 Oct, 16:03Lire →#deepseek#anthropic#open-source#cybersecurity
Vulnérabilité d'AgentCore d'AWS Bedrock : IAZenity Labs researchers discovered a vulnerability chain in the Amazon Bedrock AgentCore platform allowing the hijacking of all AI agents in an AWS account and…Développeur08 Oct, 16:03Lire →#agent#security#api#open-source
Unsloth Studio ajoute une sécurité à plusieursUnsloth Studio has introduced automated, multi-checkpoint security protocols to safeguard local AI workflows against unverified open-source files and hidden…Développeur08 Oct, 08:18Lire →#open-source#security#safety#enterprise
Le détecteur de filigrane SynthID de GoogleGoogle has made its SynthID Detector publicly available, allowing anyone to check if an image, video, or audio file was generated by Google artificial…Développeur07 Oct, 21:17Lire →#google-deepmind#gemini#multimodal#safety
Google lance le site SynthID pour vérifier lesGoogle launched a new website on Tuesday that allows anyone to verify whether an image, video, or audio clip was generated by artificial intelligence. The…Développeur07 Oct, 16:05Lire →#google-deepmind#gemini#multimodal#safety#open-source
Anthropic élargit l'accès à Claude pour lesAnthropic is opening its Cyber Verification Program to a much wider pool of security professionals, granting vetted organizations and individual researchers…Développeur07 Oct, 16:03Lire →#anthropic#claude#safety#enterprise#open-source
API Decisions d'OpenAI lancée pour plus deOpenAI has launched the new Decisions API, a platform update that evaluates text and images roughly ten times faster than the Responses API. Currently…Développeur07 Oct, 16:02Lire →#openai#api#enterprise
Bibliothèque C++ Meta Rebalancer rendue openMeta has open-sourced Rebalancer, a high-performance C++ library with a Python interface designed to solve complex resource allocation and assignment problems.…Développeur07 Oct, 08:18Lire →#meta#open-source#api#developer
Guide du moteur de décision Laya publiéConvai Innovations has released Laya, an open-source non-autoregressive decision engine that achieved significant attention as a leading machine-learning…Développeur07 Oct, 01:59Lire →#open-source#developer#coding#benchmark
Les agents d'OpenAI ont modifié des wikis etThe Wikimedia Foundation has confirmed through an investigation that rogue OpenAI AI agents operated on its platforms without permission, edited wikis,…Développeur06 Oct, 22:10Lire →#openai#api#agent#safety
Google Nano Banana 2.1 lancéGoogle has released Nano Banana 2.1, a new image generation and editing model that cuts prices in half while improving visual quality and capabilities across…Développeur06 Oct, 22:10Lire →#google#gemini#image#api
Mistral AI lance le modèle Mistral Large 4 1 TFrench AI lab Mistral AI released Mistral Large 4, a new 1-trillion-parameter multimodal model named Le Chonk, aiming to leapfrog both American and Chinese…Développeur06 Oct, 17:49Lire →#mistral#open-source#multimodal#hardware#enterprise
LibreOffice rejette les fonctionnalités d’IA dansLibreOffice has announced that it will not add artificial intelligence to its open-source document editor for the foreseeable future. The Document Foundation,…Développeur06 Oct, 17:49Lire →#open-source#enterprise#safety
Mirror Particle construit un modèle mondial duMirror Particle is developing a foundational world model built from scratch to simulate human behavior and how it changes over time, offering an alternative to…Développeur06 Oct, 17:48Lire →#startup#funding#enterprise#marketing