Key Highlights

  • Anthropic launched Claude Opus 5, which it says is its best-performing and most cost-effective model to date.
  • Opus 5 outperforms Fable 5 on coding and knowledge work evaluations at half the price.
  • Opus 5 is priced at 5 dollars per million input tokens and 25 dollars per million output tokens.
  • Anthropic said Opus 5 is not state-of-the-art for risky, dual-use capabilities such as cybersecurity.

A New Model Built Around Cost Efficiency

Anthropic announced its latest AI model, Claude Opus 5, describing it as the company's best-performing and most cost-effective offering across a number of industry benchmarks. The model outperforms Anthropic's previous release, Claude Fable 5, on coding and knowledge work evaluations, with the company stating it was designed to be used every day, according to a company blog post.

Opus 5 is priced at half the level of Fable 5, costing users 5 dollars per million input tokens and 25 dollars per million output tokens, a pricing structure that reflects the company's stated emphasis on value alongside performance. Anthropic said Opus 5 is not the state-of-the-art model for risky, dual-use capabilities such as cybersecurity, distinguishing its intended use case from other models in the company's lineup.

Responding to a More Cost-Conscious Enterprise Market

The model launch arrives as Anthropic and its chief rival OpenAI work to address a customer base that has grown increasingly focused on cost, with enterprises less inclined to experiment with pricier models absent a clear picture of return on investment. Anthropic also faces competitive pressure from rivals touting cheaper offerings, including Microsoft, Amazon, Google and a number of Chinese startups.

Dianne Penn, Anthropic's head of product management for research, told CNBC that enterprise customers are looking for value in feedback the company has received, noting that a cheaper model or offering is not actually useful if it fails to accomplish a similar level of quality compared with pricier alternatives.

Built on a Foundation Shaped by Recent Export Control Scrutiny

Opus 5 arrives against the backdrop of intense scrutiny Anthropic has faced since releasing a model called Claude Mythos Preview in April to a select group of users, which the company said was extremely capable in cybersecurity and excelled at discovering and exploiting software vulnerabilities. In June, Anthropic released an updated version, Mythos 5, alongside a separate version called Fable 5 intended to be safe for general use, with the company stating Fable 5's capabilities exceeded those of all other models it had previously released publicly.

Anthropic had to briefly disable access to both Mythos 5 and Fable 5 days after their release in order to comply with a government export control directive citing national security authorities, restoring access after roughly two weeks of intense negotiations once the controls were lifted. Penn said Anthropic continues to collaborate with government agencies as part of its pre-deployment and testing process, including for Opus 5, with the company noting that Opus 5 remains behind Mythos 5 specifically in offensive cybersecurity and biology research capabilities.

Conclusion

Anthropic's launch of Claude Opus 5 reflects a deliberate strategic response to an enterprise market increasingly focused on demonstrable value rather than raw capability alone, positioning the new model as a cost-effective option for everyday use cases while deliberately keeping more sensitive dual-use capabilities concentrated in its separate Mythos line. With competitive pressure building from both established rivals and lower-cost entrants, and given the recent history of export control scrutiny surrounding the company's more capable models, how Opus 5 performs commercially may offer an early signal of whether cost-focused positioning proves an effective strategy in the current AI market.