Grok 4.5 Enters Private Beta at SpaceX: Musk's AI Surges Ahead

On July 4, 2026, Elon Musk announces that Grok 4.5, the latest iteration of xAI's advanced AI model, is now in private beta at SpaceX and Tesla. The new model, based on a 1.5 trillion-parameter V9 foundation, integrates Cursor data to enhance coding and technical reasoning capabilities.

Musk's Bold Move in AI Development

Elon Musk reveals on X that Grok 4.5, with its 1.5 trillion parameters, is three times larger than the current 500 billion parameter v8-small model. This represents a 50% increase from Grok 4.4, which was released just a month ago. The integration of Cursor data, acquired through SpaceX's $60 billion purchase of Anysphere, aims to sharpen the model's performance in coding and technical tasks.

Internal Evaluations Show Promising Results

Internal evaluations at SpaceX and Tesla suggest that Grok 4.5 performs close to, or even exceeds, Opus, a leading AI model. Developer Mehul Mohan, who tested an early build, describes the experience as 'similar to Opus.' However, no public benchmark data is available yet for Grok 4.5.

Deployment Strategy and Future Plans

xAI is using SpaceX and Tesla as real-world testing environments for Grok 4.5. These companies provide challenging technical tasks that go beyond typical benchmarks. The V9 foundation model, which underlies Grok 4.5, completed training on May 26, 2026. xAI plans to release V9-based model variants monthly through the end of 2026. Musk also confirms that SpaceX will release 'completely new models trained from scratch' every month, a development cycle faster than any other lab has publicly committed to.

Ambitious Goals for Grok 5

Looking ahead, Grok 5 targets 6 to 10 trillion parameters, making it the largest model architecture ever discussed. It is currently training on Colossus 2, alongside six other concurrent training runs. The monthly V9 variants, including Grok 4.5, are stepping stones in the reinforcement learning and capability improvement process that feeds into Grok 5.

Continuous Improvement with Grok Build

The Grok Build coding harness, xAI's internal tool for evaluating and improving model performance, runs daily improvement cycles on Grok 4.5 during the beta period. This rapid iteration and deployment strategy positions xAI to outpace other frontier labs, leveraging production environments at Musk's own companies for evaluation.

References

• AI News Today July 4 2026: 15 Biggest Stories