July 2026 roundup tracks AI product and policy shifts

July 2026 opened with a busy run of AI developments led by xAI, Anthropic, OpenAI, Tesla and regulators in China and the US. On July 6, 2026, xAI’s Grok 4.5 entered private beta at SpaceX and Tesla on a 1.5T parameter V9 model, China prepared anthropomorphic AI rules taking effect July 15, Tesla brought Robotaxi to Miami without a safety monitor, and Grok Imagine Video 1.5 was described as beating rival video models at 86% lower cost.

Anthropic’s week combined product launches, research tools and policy pressure. Claude Sonnet 5 launched with $2/M input tokens, a 1M token context window and benchmark results of 63.2% versus Sonnet 4.6’s 58.1%, while California agencies received Claude access at a 50% discount. The company also restored Fable 5 after export controls were lifted, moved to close China access loopholes, launched Claude Science with 60+ tools, and announced an internal drug discovery program for neglected diseases.

OpenAI and wider market developments rounded out the period. The White House was reported to be drafting voluntary frontier model release standards with OpenAI, Google and Anthropic, while OpenAI proposed a 5% US government equity stake. Anthropic reported a $47B annualised run-rate compared with OpenAI’s $25-33B, and global VC reached $510B in H1 2026, with OpenAI and Anthropic taking 43%.