Elon Musk Calls Claude Opus 5.5 a 'Masterpiece' and Acknowledges Grok 4.8's Lag

Elon Musk, founder of xAI, acknowledged that the AI model Grok 4.8 being developed will fall short of Claude Opus 5.5 from the competing company Anthropic. The entrepreneur announced this on November 10, 2026, in response to a question from a user on the social network X.

"Opus 5.5 is a masterpiece," wrote Musk, calling the Anthropic model a masterpiece. In response to a question about comparison with Grok 4.8, he described his own development as "Not as good."

However, Musk expressed confidence that the next generation of his AI model will be able to surpass Claude Opus 5.5. According to his estimate, Grok 5 could reach this level within two to three months. The entrepreneur did not specify a release date for the new model.

What is known about the performance of Claude Opus 5.5

Anthropic introduced Claude Opus 5.5 on September 22, 2026. The model is designed for programming, handling large volumes of information, and performing complex multi-step tasks.

According to Anthropic, Opus 5.5 is comparable to the more expensive Claude Fable 5.1 in terms of task performance. However, the operating costs of the new version under typical load are about 40% lower than the previous Claude Opus 5.

The developer also reported improvements in programming test results and AI agent performance. However, the results of individual tests depend on the configuration and evaluation methodology used, so they do not allow for a definitive determination of the model's superiority in all scenarios.

Comparable test results for Grok 4.8 have not yet been published by xAI. Thus, Musk's assessment remains a personal comparison rather than confirmed independent testing.

xAI completes development of Grok 4.8

According to Musk, the preliminary training of Grok 4.8 has already been completed. The model is currently undergoing reinforcement learning, during which developers are improving its behavior and task performance results.

The xAI team also redesigned the software infrastructure, moving away from JAX and SGLang in favor of their own system in C and C++ languages. Additionally, developers paid extra attention to data cleaning for controlled model retraining.

Musk described the current state of Grok 4.8 as not outstanding yet, while noting improvements in its technical foundation. He expects to open limited test access for the first users within a week.

Despite the positive assessment of Claude Opus 5.5, Musk continues to count on further progress from xAI. However, the timing of Grok 5's appearance and its actual superiority over competitors have not yet been confirmed by test results.