Claude Opus 5 Rivals GPT-5, Dominates Coding Tasks

Anthropic put out Claude Opus 5 on Thursday, its first new flagship release since the company's latest standoff with Washington over model safety. The timing isn't subtle. Days earlier, an OpenAI security incident had chewed up most of the industry's attention, and Anthropic's own Fable 5 model had spent weeks offline after drawing government scrutiny over cybersecurity risks. That model, part of Anthropic's Mythos-class lineup meant for public use, came back online eventually, but only with tighter safeguards bolted on.

Opus 5 inherits those lessons directly. Anthropic says the new model carries more cyber protections than the previous Opus version, a direct response to the concerns that got Fable 5 pulled in the first place. The company is also positioning Opus 5 as a serious upgrade for coding work, claiming it approaches Fable 5's capabilities in several areas while outperforming it on complex programming tasks. That comparison matters given what Fable 5 triggered: weeks of negotiation with federal officials that reshaped how the Trump administration approaches AI oversight, and pushed OpenAI toward a similarly cautious rollout for GPT-5.6.