OpenAI. OpenAI publicly releases GPT-5.6 Sol, Terra, and Luna models across ChatGPT, API, and Codex, marking a major frontier upgrade after a government-coordinated preview delay[2][3]. OpenAI. OpenAI launches GPT-Live (GPT‑Live‑1 and GPT‑Live‑1 mini), a full‑duplex voice AI that enables natural, real-time conversational interaction and begins global rollout to ChatGPT users[2][8]. Anthropic. U.S. government lifts restrictions on Anthropic’s latest Claude versions, ending a weeks‑long cybersecurity‑related ban and restoring global access to its frontier chatbot[6][14]. Meta. Meta unveils new generative image models—Muse Image and upgraded Muse Spark 1.1—alongside opening a public preview of its Model API, pushing deeper AI integration into Instagram, WhatsApp, and developer tooling[2][5][14]. Google DeepMind. DeepMind’s “Emergent Planning in LLMs” study shows large language models can exhibit coherent multi‑step planning behavior without explicit planning architectures, highlighting a significant research breakthrough in AI reasoning[11].