Elon Musk said SpaceXAI plans to release Grok 4.6 around August 7, with Grok 4.7 expected several weeks later. The update points to two new model launches as SpaceXAI continues to compete with OpenAI, Anthropic, and China’s Moonshot AI.

Musk disclosed the plan in response to Vercel CEO Guillermo Rauch. He described Grok 4.6 as a 1.5T model with significant improvements in supervised fine-tuning and reinforcement learning.

Interesting. Grok 4.6 releases around August 7. This will be the 1.5T model with significantly improved SFT & RL. Grok 4.7 will be the 2.1T model released a few weeks later. This will be better than 4.6 in every way, except slightly slower to serve, albeit with even better… — Elon Musk (@elonmusk) July 28, 2026

Musk posted the same message on X, saying Grok 4.6 would arrive around August 7 and that Grok 4.7 would be a 2.1T model released a few weeks later. He said Grok 4.7 would be better than Grok 4.6 in every way except for being slightly slower to serve, while offering even better token efficiency.

Musk Highlights Grok’s Progress in the AI Race

Musk said the August 7 timing was only an estimate. He did not specify a launch date for Grok 4.7, and he did not provide pricing or access details in the announcement.

The remarks followed a benchmark by Rauch of Grok 4.5. In his cybersecurity benchmark, Rauch said Grok 4.5 was the strongest option when price and performance were considered together.

Rauch said Grok 4.5 was ten times cheaper than Sol by OpenAI. He also estimated that it cost 5.7 times less than Opus 5 by Anthropic and 2.2 times less than Kimi K3 by Moonshot AI.

In the benchmark, Grok 4.5 had almost the same performance as Kimi K3. Rauch, however, ranked Sol as the best model and said Sol was still better than Opus 5.

SpaceX AI launched Grok 4.5 on July 16. The company introduced it as a model for coding, agent-based tasks, and knowledge work. Grok 4.5 is estimated to cost $2 per million input tokens and $6 per million output tokens.

The latest announcements come amid intensifying competition between U.S. and Chinese AI developers, where launch timing, model size, and cost efficiency are now part of the broader product race. Moonshot AI launched Kimi K3 on July 16 as an open-weight model with 2.8 trillion parameters.

Washington Reviews Open-Weight AI Models

According to Axios, the Trump administration is considering limits on some advanced Chinese AI models, including Kimi K3.

In response, Nvidia, Microsoft, Meta and others signed an open letter opposing broad bans on open-weight AI. The signatories said open models support innovation and U.S. leadership in the sector.

Musk backed the letter. OpenAI CEO Sam Altman also supported U.S. leadership in both open-source and proprietary AI models. OpenAI, which was not on the initial list, later joined the letter as a signatory.

OpenAI and Anthropic have also called for a consistent review process for models with advanced cybersecurity and national security features. According to a report, such models can be reviewed by the federal department up to 30 days before release.