On July 24, 2026, Anthropic introduced Claude Opus 5, claiming it to be its best-performing and most cost-effective AI model to date. Priced at $5 per million input tokens and $25 per million output tokens, Opus 5 is half the cost of the previous model, Claude Fable 5, and demonstrates superior performance in coding and knowledge work evaluations.
This launch comes at a time when enterprises are increasingly cautious about investing in AI technologies without clear returns, prompting Anthropic to focus on delivering value. Dianne Penn, Anthropic's head of product management for research, emphasized the importance of balancing cost with quality, noting that cheaper models must still meet performance standards to be useful.
The company has faced scrutiny following the release of its earlier models, particularly Claude Mythos Preview, which was temporarily disabled due to government export controls.
Despite these challenges, Anthropic continues to collaborate with government agencies on the deployment of Opus 5, although it acknowledges that this model does not match the capabilities of its more advanced Mythos 5 in areas like offensive cybersecurity.
As competition intensifies from major players like Microsoft, Amazon, and Google, Anthropic's strategic focus on affordability and performance may help it capture a larger share of the enterprise AI market