Elon Musk has shared fresh details about the next stage of xAI\u2019s artificial intelligence roadmap, revealing that Grok 4.5 has entered private beta testing at SpaceX and Tesla while suggesting the company is rapidly closing the gap with the industry\u2019s leading AI models.<\/p>\n
In a series of posts on X, Musk said the latest version of Grok is built on the company\u2019s new 1.5-trillion-parameter V9 foundation model and incorporates supplementary training using data from AI coding platform Cursor. According to him, initial internal evaluations indicate the model performs at a level comparable to Anthropic\u2019s Claude Opus, with the possibility that it could outperform it in some areas.<\/p>\n
STORY CONTINUES BELOW THIS AD<\/p>\n
Grok 4.5, based on our 1.5T V9 foundation model, with Cursor data added in supplemental training, is now in private beta at SpaceX & Tesla. Early evals show performance close to, perhaps exceeding Opus.<\/p>\n
RL is continuing to significantly improve the model, and the Grok Build\u2026<\/p>\n
\u2014 Elon Musk (@elonmusk) June 28, 2026<\/a><\/p>\n The update comes as competition among AI developers continues to intensify, with companies including OpenAI, Anthropic, Google and xAI racing to release increasingly capable large language models while investing heavily in computing infrastructure.<\/p>\n Grok roadmap accelerates<\/p>\n Musk said reinforcement learning continues to improve Grok\u2019s capabilities, while the company\u2019s development tools are becoming more efficient. He added that SpaceX plans to release completely new AI models trained from scratch every month for the rest of the year, signalling a significantly faster development cycle than in previous generations.<\/p>\n Despite his optimistic assessment, Musk also sought to temper expectations. Responding to another X user, he clarified that he was not claiming the V9 foundation model would dramatically outperform every competing system. Instead, he described it as \u201ca solid workhorse in the same league as Opus,\u201d indicating that xAI\u2019s immediate goal is competitive parity rather than a clear lead.<\/p>\n To be clear, I\u2019m not saying the Grok v9 foundation model will be mind-blowingly better than anything, but it will be a solid workhorse in the same league as Opus.<\/p>\n And the SpaceXAI cadence of model and harness improvement is speeding up tremendously, particularly due to a few\u2026<\/p>\n