Contenu long format, mises à jour produit et notes de recherche.
Alibaba Cloud just launched an AI Coding Plan starting at $3/month that bundles Qwen, Kimi, and other models with direct integration into Cline and Claude Code. For developers tire...
Anthropic started rolling out voice mode in Claude Code to ~5% of users, with broader availability coming. For developers, this opens up hands-free coding scenarios—from quick edit...
OpenAI is rolling out GPT-5.3-Codex-Spark to engaged Codex subscribers on ChatGPT Plus, promising speeds over 1,000 tokens per second. For developers, this could mean near-instant...
Tongyi Lab open-sourced CoPaw, their personal AI partner platform, after a complete engine overhaul. The focus: model freedom and local-first architecture for users who want AI ass...
xAI's latest Grok update focuses on precision—better instruction following, reduced hallucinations, improved LaTeX rendering, and more reliable image search. For technical users, t...
Ollama now supports the full Qwen 3.5 small model series locally, including tool calling, reasoning, and multimodal capabilities. For developers building AI-powered applications, t...
Alibaba's Qwen team released a family of small language models—from 0.8B to 9B parameters—that deliver multimodal capabilities and tool use on modest hardware. For businesses, this...
Zopia AI launched an end-to-end video director agent that orchestrates multiple AI models to create cinematic content from a single concept. For content teams, this could mean prod...
Browser Use just shipped autonomous AI agents that don't need constant prompting—they run continuously with scoped authentication, personal workspaces, and integrations to Slack, L...
Google Translate is rolling out AI-powered updates that understand context and nuance—meaning your international business emails and documents will finally sound like they were wri...
Google Workspace is shipping cartoon avatars for Google Vids and replacing the original Gemini app model with Nano Banana 2—promising dramatically faster AI responses without sacri...
Ollama's OpenCode now supports subagents—letting you parallelize complex tasks like research, refactoring, and code reviews by spinning up multiple AI agents that work simultaneous...
Anthropic just democratized AI tool integration—150+ connectors for coding, design, finance, and sales are now available on Claude's free plan.
Figma is rolling out Nano Banana 2, an AI image generator that promises faster outputs and professional-grade results directly inside your design workflow.
Google's latest image model combines the quality of Pro with the speed of Flash—giving marketers and designers a viable alternative to Midjourney and DALL-E.
A new open-source approach lets you automate Meta ad management without paying for expensive SaaS tools—just API costs and your own infrastructure.
Anthropic just added Remote Control to Claude Code, letting you start a task on your laptop and manage it from your phone while you're in a meeting. This isn't just convenience—it'...
@cursor_ai shared a product update on X. This article explains what changed and what it means in practice.
Cognition's Devin 2.2 just dropped, and it's addressing the biggest pain point in AI coding: reliability. While other AI assistants write code and hope it works, Devin now tests wi...
Google just acquired ProducerAI (formerly Riffusion) and folded it into Google Labs, and this is a direct shot at Suno and Udio. The music AI space has been dominated by startups;...
Google just added animated SVG generation to Gemini 3.1 Pro, and it's a bigger deal than it sounds. For marketing teams and product designers, this means custom animations without...
Google Labs just upgraded Opal with a new agent step that analyzes your goal, picks the right tools, and executes autonomously. This is the difference between 'AI-assisted' and 'AI...
@GoogleLabs shared a product update on X. This article explains what changed and what it means in practice.
OpenAI just made GPT-5.3-Codex available to all developers via the Responses API, and this is bigger than it sounds. It's not just a new model—it's OpenAI's bet that coding agents...
Pencil just crossed 100,000 users and launched SWARM mode—a multi-agent system where several AI designers work in parallel on your project. This isn't just another AI design tool;...
Perplexity and Comet rolled out voice mode upgrades today, and the timing isn't accidental. With ChatGPT's Advanced Voice Mode winning users and Gemini Live feeling half-baked, Per...
ByteDance's BytePlus just dropped Seedream 5.0 Lite, and it's not just another Midjourney clone. This thing understands spatial relationships, design intent, and can work from sket...
Guillermo Rauch just announced that all designers at Vercel now build production code using v0, Claude Code, and Cursor. This isn't a cute experiment—it's a fundamental restructuri...
CastFox has integrated Gemma 3n via Ollama to process podcast audio in English, Japanese, and Korean. The implementation enables contextual chat, smart highlights, and semantic sea...
Cursor has released a major upgrade introducing cloud-hosted AI agents. These agents can autonomously onboard to codebases, operate remote desktop environments, and generate video...
Google DeepMind has unveiled Nano Banana 2, a new image generation model powered by Gemini 3.1 Flash. Early testing shows output quality rivaling Nano Banana Pro at a lower cost an...
Notion has launched Custom Agents, making AI automation accessible to businesses that lack dedicated AI teams or forward-deployed engineers. This represents a significant democrati...
Perplexity and Comet are rolling out upgraded voice mode capabilities to all users. The new features include screen-aware queries, site navigation via voice, and persistent context...
