Anthropic released its newest model, Opus 5.5, on Tuesday, claiming a new state of the art in coding and knowledge work performance. Opus is the most capable and most expensive tier in Anthropic's three-tier Claude lineup; Sonnet sits in the middle, while Haiku is the fastest and cheapest. According to Anthropic, the release surpasses the larger Fable model on many benchmarks and completed a number of informal tasks that Fable could not finish.
The model also costs significantly less to run than its predecessor. Anthropic will charge $20 per million output tokens for Opus 5.5, down from $25 for the previous model, with other metrics showing similar price reductions. The model is also faster, reflecting an overall drop in the compute required to serve it. In addition, Opus 5.5 changes how it communicates, using less jargon and placing important information at the start of its messages.
The launch comes just two months after Opus 5 was released on July 24. According to the announcement, Sonnet 5.5 and Haiku 5.5, the next tiers in the lineup, will arrive in the coming weeks with similar performance improvements. Anthropic says Opus 5.5 is comparable to Mythos in biology and cybersecurity capabilities, so it falls under the same safeguards as the company's Fable model. Those safeguards limit how much the models can be used to find exploits in compiled programs or to develop recognizable biological weapons, among other tasks.
Opus 5.5 is Anthropic's first model release since CEO Dario Amodei embraced calls to pace the frontier, deliberately slowing progress on AI capabilities to match the pace of alignment work. In a post earlier this month, Amodei wrote that he had become convinced that fully addressing the risks requires even more prudence, including not just investing in risk prevention but pacing the rate of capabilities advancement so that risk prevention has time to keep up.
Safety training for Opus 5.5 was broadly similar to that of its predecessors, with alignment testing and pre-release evaluation by outside organizations including METR and Frontier Design. Anthropic emphasized that more advanced training and evaluation systems are already being prepared for future models, including improved security and monitoring systems. The company said in its blog post that as AI becomes more capable, public policy should play a larger role in ensuring the systems people rely on are safe. That capacity takes time to build, according to the post, and Anthropic said it has started putting infrastructure in place to support it and expects to share more details soon.
More AI news from TechManNews.
