Anthropic has released Claude Opus 5, its latest artificial intelligence model built for coding, professional work and complex "agent-based" tasks that let software carry out multi-step jobs on a user's behalf. The launch positions the company as a direct competitor to OpenAI's GPT-5.6.
According to Briefs Finance, the new model cuts costs in half while boosting everyday performance — a combination that matters because the price of running AI has been one of the biggest barriers to wider business adoption.
For existing customers, the rollout changes what they get by default. As reported by Digit, Opus 5 is now the default model on the Claude Max subscription and the strongest option available on Claude Pro, positioned as a step up from the earlier Sonnet 5 model.
The release also surfaces a sharper safety question. Tech Times, citing the model's safety card, reports that Claude Opus 5 successfully hacked enterprise networks in 8 of 10 government tests. Figures like that cut both ways: they demonstrate how capable the model has become at security-relevant tasks, while underlining why AI developers publish safety documentation and why regulators and enterprises are watching offensive capabilities closely.
The broader picture is an accelerating race at the top of the AI market. With Anthropic shipping Opus 5 as a rival to GPT-5.6, the leading labs are now leapfrogging one another on capability, price and the ability to act autonomously rather than just answer questions.
Why it matters: cheaper, more capable AI models that can both do professional work and probe computer networks bring the technology's economic promise and its security risks into the same release.