Anthropic isn’t just widely thought to have the best coding models, but it also seems to have a fan at a rival lab.
Elon Musk has said that Anthropic has done something “quite special” with coding. Musk was responding to a comment to an X user who had been speculating on how Musk’s Grok 4.20 would fare against Anthropic’s Opus 4.5. “Grok might do better with v4.20. We shall see,” Musk said.
Musk then clarified what he meant. “Not quite on programming, but it will excel in other areas. Anthropic has done something special with coding. It was a helpful motivator that they cut us of xAI and not good for their karma,” Musk said.
Musk has previously praised Anthropic’s models. “I must give @AnthropicAI credit here: Opus4.5 is outstanding,” he’d posted on X last month. However, last week Anthropic had cut off xAI’s access to its models, which the teams had been using internally to write code. Musk acknowledged that this had happened, and said that the block gave xAI further incentive to quickly catch up with Anthropic’s ability at coding.
The acknowledgment from Musk represents a significant moment in the increasingly competitive AI landscape, where even fierce rivals are compelled to recognize genuine technical breakthroughs. Anthropic’s coding capabilities have clearly set a benchmark that other labs are racing to match, and Musk’s candid assessment suggests the gap may be wider than many industry observers expected. While xAI has made impressive strides across multiple domains, Musk’s transparency about Anthropic’s coding edge reveals the highly specialized nature of modern AI development—where dominance in one area doesn’t automatically translate to leadership across all capabilities.
The decision by Anthropic to cut off xAI’s access to its models, while potentially contentious, underscores the strategic value of these coding capabilities and the intensifying competition in the AI sector. As these companies push boundaries in different directions—with Grok 4.20 expected to excel in areas beyond programming—the industry may be entering an era of specialized excellence rather than all-purpose supremacy. For enterprises evaluating AI coding assistants, Musk’s assessment offers valuable guidance: Anthropic’s models currently represent the gold standard for programming tasks, even as competitors rapidly innovate in pursuit of that “quite special” edge.