Model launchesXi Jinping Champions Open-Source AI as China Seeks Global LeadershipChinese leader Xi Jinping used his first appearance at the World Artificial Intelligence Conference to promote open-source AI and position China as a contender to reshape the global AI order.2 min read
Model launchesOpenAI launches GPT-Live voice models that listen and speak continuously while outsourcing hard reasoning to GPT-5.5On July 8 OpenAI released GPT-Live-1 and GPT-Live-1 mini, new full‑duplex voice models that can process incoming and outgoing audio simultaneously and hand complex queries to a background reasoning model (GPT‑5.5).5 min read
Model launchesThinking Machines Lab publishes Inkling, an Apache-2.0 open-weights multimodal modelThinking Machines Lab, led by Mira Murati, released Inkling, an Apache‑2.0 licensed multimodal Mixture‑of‑Experts transformer with 975 billion parameters (41 billion active) trained on roughly 45 trillion tokens.3 min read
Model launchesApple gains Chinese approval for Apple Intelligence after Alibaba and Baidu collaborationsChina’s Cyberspace Administration has approved Apple Intelligence following agreements to integrate Alibaba’s Qwen model into Apple’s operating systems and cooperation with Baidu on local features.2 min read
Model launchesGPT-Live development: real-time, multitask conversation and itinerary planningA model called GPT-Live can continuously maintain a conversation while performing multiple tasks simultaneously, such as checking flights, displaying local weather, and creating real-time itineraries.1 min read
Model launchesCohere says AI sovereignty requires control of the entire agent stackAt VB Transform 2026 in Menlo Park, Cohere VP Rachad Alao argued that corporate AI sovereignty means controlling data, infrastructure and the full agent stack rather than merely downloading an open model.4 min read
Model launchesOpenAI's Altman warns GPT-5.6 'Sol' may face short-term scaling hiccupsOpenAI CEO Sam Altman cautioned that the recently released GPT-5.6 Sol model could encounter short-term scaling problems as demand surges.2 min read
Model launchesAnthropic's Fable 5 and OpenAI's GPT-5.6: simultaneous launches, export controls and who drew more search interestAnthropic and OpenAI launched their most capable models within weeks of each other, but both faced U.S.3 min read
Model launchesMeta unveils Muse Image and Muse Spark 1.1 amid privacy concerns and developer competitionMeta introduced two new AI models this week: Muse Image, an image-generation tool that can use photos from public Instagram accounts without explicit permission, and Muse Spark 1.1, an agentic coding model with competitively priced plans.3 min read
Model launchesCritics Demand Consent After Meta Lets Users Generate AI Images Using Public Instagram LikenessesMeta introduced Muse Image within Meta AI, allowing users to generate images using the likenesses of adults with public Instagram accounts unless those people opt out.3 min read
Model launchesOpenAI consolidates GPT-5.6 features into a single desktop app, retires standalone toolsOpenAI released GPT-5.6 and on the same day consolidated its separate desktop tools into one unified application.2 min read
Model launchesGoogle launches Nano Banana 2 Lite image model and Gemini Omni Flash video model for developersGoogle has made Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) and the multimodal Gemini Omni Flash video model available to developers via its APIs and platforms.4 min read
