Model releases, breakout repos, app launches, pricing shifts, security stories, and funding rounds from across the AI ecosystem. We curate the news for non-techies, vibe builders, and developers, with a clear take on what each story means for them. Updated daily.
Two compact models hit Hugging Face trends while Vercel removed data download fees and Moonshot AI shipped a new Kimi version that stirred debate.
Lovable updates its chat interface to render generated or referenced files as visual thumbnails and clickable tiles instead of plain paths.
Lovable introduces agent integrations and Model Context Protocol links that let AI assistants call actions from published apps.
Lovable apps now automatically recognize the signed-in user without a separate login page, reading their name, email, and ID directly.
Admins can now configure and distribute Team MCP servers across cloud agents, the IDE, and CLI, and restrict marketplace access to specific organization groups.
Cursor adds side chats that run alongside main chats. It also adds local search of agent transcripts with Cmd+K and redesigned project and repo pickers.
Vendors pushed model access, CLI plugins, and new Replicate tools while safety issues surfaced with GPT-5.6 and MCP setups gained traction.
The AWS ML Blog published a post on Amazon Quick as an agentic AI tool for sales organizations.
OpenRouter launches Auto Router (Beta). The task-aware router classifies requests and routes each one to the most popular model for that task based on aggregate spend.
Claude Code pushed dozens of CLI fixes and defaults this week while Cursor shipped five releases with automation and mobile agents and Codex added plugins plus desktop integration.
Lovable now renders generated or referenced files as clickable thumbnail tiles in chat. Images and videos display visual previews, while other files appear as labeled tiles that open in the Files panel.
Lovable converts published app functionality into actions that AI assistants like ChatGPT and Claude can call. It proposes actions, manages permissions, and generates an MCP link.
Lovable now lets built apps recognize signed-in users automatically by reading name, email, and ID. The feature is available on Business and Enterprise plans for newer TanStack Start projects.
Hugging Face models and Google updates lead releases today while Vercel, NVIDIA, and new agent startups add practical tools for inference, robotics, and business workflows.
Kimi launches K3, a multimodal open-weight model with 2.8 trillion parameters and 1M token context. It nears GPT-5.6 Sol and Claude Fable 5 in benchmarks, with full weights due by July 27.
Google adds personalized AI avatars to Vids that let users create videos starring digital versions of themselves, alongside Gemini Omni-powered tools for generating and editing videos from prompts and reference images.
Meta's Muse Spark 1.1 multimodal reasoning model is now available on OpenRouter with 1,049k context at $1.25/M input and $4.25/M output.
OpenAI released ChatGPT basketball as its first hardware product this week.
Perplexity introduces secure sandboxes for agents that provide isolated environments to safely execute code and run complex tasks.
Lovable restores the Edit with Lovable badge on the next project publish after a workspace downgrades to a plan without badge-hiding support. Live sites remain unaffected until republished.
Lovable sends email notifications to workspace admins and owners about role changes and member removals. The platform shows role details and requires confirmation before promotions.
Thinking Machines Inkling model expands across gateways and platforms while new image generators and no-code agent tools appear on Replicate and LangChain.
AWS ML Blog published a post on building visual intelligence using Amazon Bedrock and MCP servers.
Thinking Machines Lab released Inkling, a 975-billion-parameter open source model trained to understand video and audio.
Thinking Machines releases Inkling after a year and a half building AI infrastructure privately.
