Anthropic, a leading AI startup competing with OpenAI and Google, has introduced Claude Opus 5, a new AI model that offers comparable performance to its most advanced system at half the cost.

The model excels in coding and knowledge work while using fewer computational resources, addressing concerns about the high expenses of running advanced AI systems.Opus 5 costs $5 per million input tokens and $25 per million output tokens, significantly lower than its predecessor, Fable 5.The company emphasizes that Opus 5 matches Fable 5's capabilities across various tasks despite the reduced cost.

Anthropic also clarified that Opus 5 is not trained on cybersecurity tasks, maintaining stricter controls over its most powerful system, Mythos 5, which handles offensive cyber operations and biological research.The new model's safety filters are expected to intervene 85% less frequently than Fable 5, reducing risks.This development comes as Anthropic and OpenAI prepare for stock market listings, drawing closer scrutiny from Wall Street.The article highlights the growing importance of cost efficiency in AI technology and the industry's efforts to balance innovation with security.