Elon Musk offered rare praise to a competitor on X this week, and he also gave a blunt answer about his own company’s next model. When a user named Dimdv asked how Grok 4.8 would compare with Anthropic’s Claude Opus 5.5, Musk said it would be “not as good.” He then called Opus 5.5 a “masterpiece.”
Why Grok 4.8 is late
Grok 4.8 is the update Musk expects to put in some users’ hands this week. It has finished pretraining, but Musk said it is still going through reinforcement learning. That phase uses feedback and rewards to refine responses. He said the team also spent a lot of time cleaning supervised fine-tuning data, the examples that teach a model to follow instructions.
Musk said xAI dropped JAX and SGLang. The team moved to an inference and reinforcement-learning system rewritten in C and C++, with Musk describing it as “C with a little C++” after a developer celebrated the change. Inference is the step that generates answers after a prompt. Rewriting that software can give engineers more control, though Musk did not share numbers showing a gain. He also credited Starlink engineers and pointed to SpaceX’s habit of writing its own headers and handling precise time calculations, including leap seconds.
Musk called Grok 4.8 “not yet incredible,” but said its foundations are finally built correctly. A soft rollout to Bot users is expected later this week. That will give xAI an early test before a wider release.
Anthropic’s lead is real
Anthropic released Opus 5.5 on September 22. The company says the model matches its higher-end Fable 5.1 on most work while costing 40% less to run than the previous Opus version on typical workloads. It is aimed at hard programming jobs and multi-step agent tasks with limited human help.
On Terminal-Bench 4.0, which tests whether agents can finish complex command-line tasks, Anthropic reported 66.4% for Opus 5.5. OpenAI’s GPT-6 Astra scored 57.9%. Grok is far behind on the leaderboards, and Musk has not published Grok 4.8 scores that would show the size of the gap.
The Grok 5 bet
Musk says Grok 5, the model after this update, “should exceed Opus 5.5.” He put the timing at perhaps two or three months. He did not give a firm release date or more details. Musk has a history of missing his own target dates, so the claim is best treated as a goal, not a schedule.
That makes this week’s limited rollout a small but important step. Grok 4.8 may not beat Claude Opus 5.5. Musk seems to know that. The question is whether the work underneath, including the C and C++ rewrite, helps xAI close the distance when Grok 5 arrives.
