Meta launches AI-powered Meta Glasses in partnership with EssilorLuxottica, featuring smart audio, hands-free capture, and 26 frame options.

Google Gemini’s macOS app is testing system-wide voice dictation, cursor tracking, and possible device linking.

DeepReinforce open-sourced Ornith-1.0, a coding model family that writes its own RL scaffolds and matches Claude Opus 4.7 on benchmarks.

Cursor launches a native iOS app, now in public beta for paid plans, that lets developers review and manage coding agents on their iPhones.

Meituan’s LongCat-2.0 public release offers a 1 million-token context window, a Mixture-of-Experts system, and API access with flexible billing.

Bloome brings people and AI agents into one chat, where agents draft, challenge, and cross-check each other before delivering a result.

Apify integrates with Coinbase’s x402 protocol, letting AI agents pay in USDC and run 20,000+ web automation Actors without accounts or API keys.

NoimosAI launches its Creative Agent, generating brand assets based on data-backed patterns from top-performing market creatives.

What's new? Anthropic releases Claude Sonnet 5 for developers on multiple plans via API; it plans tasks and uses browsers and terminals with cyber safeguards;

Anthropic prepares to restore Claude Fable 5 in the Claude iOS app, hinting at credit-gated access and identity checks after its suspension.

What's new? Nano Banana 2 Lite is a cheap text-image model via Gemini API on Search and Photos; Gemini Omni Flash is in preview for 10s video generation from text, images and clips;

What's new? Anthropic's Claude Science is an AI-powered research workbench for life sciences with 60+ tools; it works on macOS, Linux, local machines, remote clusters, and SSH;

Vellum introduces agent-to-agent communication in Slack, letting each user’s assistant collaborate directly with others while maintaining personal context.

Condense.chat's proxy compresses coding-agent context with two in-house models, cutting token bills by up to 72 percent on deep sessions.

xAI launches Voice Agent Builder beta, letting teams create custom phone agents in minutes without coding, with 80+ voices and real-time tools.

Mistral AI introduces Leanstral 1.5, an open-source code agent for Lean 4 formal proof engineering, now available via Labs API and Hugging Face.

ByteDance’s Dreamina Seedance 2.5 is set to launch in July, bringing up to 3-minute AI video generation to creators.

References to Grok 4.5, built on a new V9 foundation, are surfacing on the Grok web, hinting at a public release for Heavy subscribers soon.

xAI rolls out 21 new Grok Voice flagship voices, boosting multilingual support and production-ready voice agent creation for developers and businesses.