Model launchesOpenAI removes chat limits for free ChatGPT users with GPT-5.6 Luna rolloutOpenAI is updating the language models behind ChatGPT: free and Go users will receive GPT-5.6 Luna, which removes the text-chat message limit and adds a "Think" button that engages deeper reasoning.2 min read
Model launchesLiquid AI releases LFM2.5-2.6B: an agent-focused, local-first 2.6B-parameter modelLiquid AI, founded in 2023 by former MIT computer scientists, has released LFM2.5-2.6B: a 2.6-billion-parameter, open-weight language model designed for agentic workloads and local execution from Raspberry Pi to servers.7 min read
Model launchesUpdated GPT‑5.6 Sol version available in the ChatGPT Chat experience from todayChatGPT Plus and Pro subscribers can access the updated GPT‑5.6 Sol version on the Chat interface from today, which is intended for everyday conversations.1 min read
Model launchesLeading AI Researchers Say We May Be at the Threshold of a 'Singularity'Senior researchers and executives at major AI labs increasingly assert that developments in frontier models indicate the beginnings of a machine-driven ‘intelligence explosion’.4 min read
Model launchesMeta launches Muse Code and Muse Spark 1.2 with emphasis on long-sequence coding and tool callingMeta released two models — Muse Code and Muse Spark 1.2 — focused on improved code generation, long-horizon workflows and agentic tool calling.2 min read
Model launchesHark launches Handoff agent for autonomous web tasks, claims top benchmark score and lower pricingHark unveiled Handoff, a "computer use agent" that autonomously navigates the open web to complete tasks like ordering food or booking flights.5 min read
Model launchesOpenAI’s ChatGPT Work: a cloud agent for knowledge work and how it functionsOn July 9, 2026 OpenAI launched ChatGPT Work, a cloud-based agent product aimed at knowledge workers that combines ChatGPT, Codex, browser automation and a plugin ecosystem.9 min read
Model launchesFrom VLAs to World Action Models: Cosmos 3 as a Foundation for Physical Generalization in Robot PoliciesResearchers are shifting robot policy backbones from vision-language models (VLAs) to video-based world models, producing World Action Models (WAMs) that learn dynamics as well as semantics.5 min read
Model launchesAlibaba debuts Qwen3.8‑Max, a 2.4T‑parameter MoE model aimed at autonomous software engineeringAlibaba's Qwen team introduced Qwen3.8‑Max, a 2.4‑trillion‑parameter mixture‑of‑experts multimodal model positioned for long‑horizon autonomous software engineering and enterprise automation.4 min read
Model launchesOpenAI: GPT-Live can listen and speak to the user simultaneously thanks to continuous audio architectureOpenAI announced that GPT-Live can listen and speak at the same time after the entire voice system was rebuilt from the client side to the model.1 min read
Model launchesDeepSeek launches discounted V4‑Flash as China’s AI price war intensifiesChinese AI firm DeepSeek has released an open‑weight model called V4‑Flash with improved agent capabilities and a 50% reduction in token costs, while delaying a planned dynamic‑pricing feature.2 min read
Model launchesApple ships functional Siri AI in iOS 27 beta, but arrival feels overdueApple has introduced a functional Siri AI in the iOS 27 consumer beta released in July, delivering promised capabilities like contextual understanding, access to on‑device data, and a conversational interface.3 min read
