Moonlake AI just launched the open beta for its world modeling agent, promising to build playable games up to 100x faster. Instead of static 3D scenes, it generates physics, audio, and gameplay logic unified across interactive objects. It is rolling out to over 10,000 waitlisted creators, with subscriptions starting at $15 a month.
Anthropic just launched Claude Sonnet 4.6, delivering substantial upgrades across coding, computer use, and long-horizon reasoning. In early tests, developers preferred it over Sonnet 4.5 70% of the time and even picked it over the frontier Opus 4.5 in coding tasks. Why it matters: It offers near-flagship performance at mid-tier pricing and is already the default model for Free and Pro users on claude.ai.
Anthropic just added dynamic filtering to Claude's web search and fetch tools on the API. Instead of dumping raw HTML into context, Claude writes and runs code to strip junk data before reasoning over it, boosting benchmark accuracy by 11% while slashing input tokens by 24%. It is on by default for Sonnet 4.6 and Opus 4.6.
Figma just launched a workflow with Anthropic that captures live UI from Claude Code or any browser into fully editable Figma frames. Teams can now tweak, branch, and annotate AI-generated prototypes visually without touching code. Combined with Figma's MCP server, you can push changes straight back into code. Why it matters: It bridges the gap between fast AI vibe-coding and collaborative UI design.
ElevenLabs just launched ElevenAgents for Support, conversational AI agents that handle customer service across voice, phone, email, and WhatsApp in over 90 languages. The platform boasts sub-second response times, over 10,000 expressive voices, and tone adaptation to de-escalate heated issues without awkward pauses. Teams can deploy them directly from existing SOPs with zero coding and hand off complex tickets to human reps with full context.
Cursor just launched a plugin marketplace that lets its AI agents connect directly to external developer tools. These plugins package MCP servers, skills, and subagents to handle tasks across the product cycle—like generating Stripe payments, pulling Figma designs, or deploying to AWS. Users can install community plugins or build their own, with private team hubs on the way.
Cyborg Dynamics and Griffith University just tested an AI-powered robot swarm that put out fires with a 99.67% success rate. Using multi-agent reinforcement learning across physical and virtual ground vehicles, the system self-organized to navigate obstacles and divide tasks. Why it matters: Autonomous swarming could soon replace manual remote controls and keep first responders out of extreme danger.
Google just rolled out Ideate Agent inside its experimental Stitch tool, taking design tools beyond simple UI generation into deep research. It browses the web to analyze competitor patterns and cooks up multiple design directions before you commit to a single canvas. It is currently free in beta via Google Labs.
Meta just integrated Manus AI into Ads Manager, giving all advertisers access to autonomous agents for report building and audience research. As investors grow uneasy over hundreds of billions poured into AI infrastructure, Meta is pushing to turn its shiny agent tech into immediate ad efficiencies.
Alibaba's Qwen team just rolled out Qwen3.5-397B-A17B, the first open-weight model in the Qwen3.5 lineup. Built on a sparse MoE architecture, it carries 397 billion total parameters while activating just 17 billion per token, pushing decoding throughput up to 19x over Qwen3-Max. Why it matters: Developers get Apache 2.0-licensed multimodal agent capabilities without the prohibitive serving costs of dense giants.
Manus just rolled out Manus Agents, letting users run its full multi-step AI agent directly inside Telegram via a one-minute QR code setup. It supports voice, image, and file inputs, with switches between Manus 1.6 Max and Lite models. Why it matters: Bringing full autonomous agent execution into messaging apps cuts out the friction of web dashboards, turning your daily chat into an active workspace.
OpenAI just hired Peter Steinberger, the developer behind OpenClaw, to head up its next generation of personal agents. Quoting @sama, his work will quickly become core to OpenAI's product offerings, while OpenClaw itself moves to a foundation to stay open source. Why it matters: OpenAI is aggressively betting on autonomous, multi-agent systems.
Moonshot AI just integrated the OpenClaw framework directly into kimi.com with Kimi Claw, moving complex agent workflows from local setups to a 24/7 cloud tab. Users get 40GB of storage for persistent RAG datasets, access to ClawHub's 5,000 community skills, and Telegram bridging for automated tasks.
HKU Data Intelligence Lab just launched ClawWork, an evaluation benchmark that forces AI agents to earn their own living. Agents start with just $10 and must pay for their API token costs while completing real tasks across 44 professions—if they run out of cash, they go bankrupt. It basically replaces synthetic benchmark scores with raw economic viability.
OpenClaw founder Peter Steinberger just announced he is joining OpenAI to build AI agents everyday people can use. Instead of spinning his viral project into a startup, he is moving OpenClaw to an independent foundation sponsored by OpenAI, ensuring it stays open source and model-agnostic.
Alibaba just dropped Mobile-Agent-v3.5 alongside the GUI-Owl-1.5 model family, spanning from 2B to 235B parameters. Built on Qwen3-VL, it natively unifies vision, reasoning, and tool execution across desktop, mobile, and browser environments, topping 20+ benchmarks. Model weights are now open on Hugging Face with live demos available.
Airbnb just announced plans to embed large language models across its entire app, spanning conversational search, trip planning, and host tools. A natural language search feature is already rolling out to a tiny sliver of users, while its customer support bot already resolves a third of issues without human help. Next up: conversational voice support.
ByteDance just open-sourced OpenViking under Apache 2.0, introducing a virtual file system to manage agent contexts, memories, and skills instead of relying on flat vector slices. It uses tiered loading across three summary levels to slash token overhead while fully visualizing the agent's retrieval path.
NVIDIA just dropped a guide on running the popular OpenClaw AI agent entirely for free on GeForce RTX GPUs and DGX Spark hardware. By pairing local runners like Ollama or LM Studio with open models, users can let an always-on agent manage local files and schedules without bleeding cash on cloud API tokens or handing over private data. It brings powerful autonomous desktop workflows right to consumer hardware.
Microsoft AI CEO Mustafa Suleyman just predicted that AI will achieve human-level performance on most professional tasks within the next 12 to 18 months. He expects roles like lawyers, accountants, and project managers to see the vast majority of their computer-based tasks fully automated, pointing to software engineering where AI already writes most production code.
