NEW DELHI — xAI’s latest artificial intelligence model, Grok 4.5, has entered private beta testing at SpaceX and Tesla ahead of a broader release, Elon Musk said Sunday.
In a post on X, Musk said the model is based on xAI’s 1.5-trillion-parameter V9 foundation model and received additional training using data from the AI coding assistant Cursor.
Early evaluations suggest Grok 4.5 is performing near the level of Anthropic’s Claude Opus and may surpass it in some areas, according to Musk.
“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,” Musk wrote.
Musk said reinforcement learning continues to improve the model’s capabilities. He also said xAI’s internal coding system, known as the Grok Build coding harness, is becoming more effective and thanked the teams involved in its development.
He added that xAI plans to release new AI models trained from scratch through SpaceX every month for the remainder of the year, pointing to a faster development schedule for the company’s AI technology.
Musk first disclosed the V9 foundation model in late May, saying it had completed training with 1.5 trillion parameters. The model is about three times larger than the 500-billion-parameter v8-small model currently handling Grok workloads on X.
The training process included substantial data from Cursor in an effort to strengthen Grok’s software coding capabilities.
The V9 rollout has proceeded gradually since training was completed in late May. Subsequent updates had indicated that the underlying V9-Medium model could receive a public release in mid-June.
Sunday’s announcement marked the first confirmation that a model officially branded as Grok 4.5 had advanced to private testing within Musk’s companies. (Source: IANS)
