June 30 brought a wave of model and platform shuffling: OpenAI previewed a three-tier GPT-5.6 family (Sol, Terra, Luna), Google throttled Meta's access to Gemini, and Adobe scooped up Topaz Labs to bolster its media-enhancement stack. Anthropic's continued export restrictions are reshaping the competitive map, with Chinese and other Asian labs shipping Mythos-like alternatives (including a GLM-5.2 model that reportedly beats Claude on cyber benchmarks) rather than waiting it out. Enterprise adoption data from RBC and Anthropic's own Economic Index show AI spend and usage climbing past the pilot stage, even as governance products (Okta, Workday) and red-team writeups (Simon Willison's assistant-hacking log) underline that guardrails haven't caught up. On the tooling side, GitLab's integration with Google Antigravity and Epic Games' new open-source version control system, Lore, point to infrastructure quietly adapting for agent-heavy, high-churn engineering workflows.

• OpenAI previews GPT-5.6 as three models: Sol, Terra, and Luna
• Google restricts Meta's use of its Gemini models amid rising AI rivalry
• Adobe acquires Topaz Labs to strengthen AI image/video enhancement
• Anthropic's export ban pushes Asian labs to ship Mythos-like rivals, one beating Claude on cyber tests
• GitLab Orbit pairs with Google Antigravity to give coding agents full repo context