Anthropic Launches Claude Opus 5, Undercutting Fable 5 on Price as Businesses Balk at AI Costs

Anthropic released Claude Opus 5 on Friday, positioning it as the model built for daily use rather than one chasing the highest possible benchmark score.

In its own announcement, the company said Opus 5 is a thoughtful and proactive model that comes close to the frontier intelligence of Fable 5 at half the price.

Per Anthropic, on evaluations like Frontier-Bench and GDPval-AA, it now represents the new state of the art among publicly available Claude models, trailing only the restricted Mythos on cybersecurity tasks.

Pricing Undercuts Fable 5 by Half

Anthropic’s own documentation on the Claude Platform confirms the new model launches at $5 per million input tokens and $25 per million output tokens, identical to what Opus 4.8 charged and exactly half of Fable 5’s $10-and-$50 structure. 

Opus 5 is launching only about two months after Opus 4.8, which arrived on May 28, following a busy stretch that saw Mythos 5, Fable 5, and Sonnet 5 all launch in June, leaving Haiku as the only model in the lineup still waiting for its own upgrade. 

TechCrunch also pointed out that Opus 5 actually outperforms Fable 5 on several benchmarks Anthropic included in its announcement, an unusual result for a model priced and marketed a tier below its more expensive sibling.

Businesses Have Grown Wary of Fable 5’s Costs

CNBC reported that Anthropic has had to pacify a more cost-conscious customer base, where enterprises have grown less inclined to experiment with expensive models without a clear picture of return on investment. 

That pressure has grown since Fable 5’s debut last month, with higher pricing and compute demands opening the door for cheaper alternatives like newly launched open-weight Kimi K3, a rival to Claude’s flagship models. 

CNBC added that Anthropic continues working with government agencies on pre-deployment testing, including Opus 5, after Mythos-class models were briefly removed from public access in June to comply with export controls before being restored.

Safety Guardrails Loosen as Fallback Options Expand

Beyond pricing, Anthropic also detailed changes to Opus 5’s handling of restricted requests. 

The company said its cyber classifiers are less restrictive than Fable 5’s, allowing source code vulnerability searches while blocking binary scanning, penetration testing, and exploit generation.

TechCrunch reported that Anthropic is also introducing Automatic Fallbacks, a beta feature that routes flagged requests to a less restricted AI model, giving API users a response instead of a refusal. 

Like Opus 4.8 before it, Opus 5 remains free of the 30-day data retention requirement applied to Fable 5 and Mythos 5, avoiding the same data storage concerns, a reason which could make Opus 5 more suitable for everyday business use despite being a lower-tier model. 

Source: Introducing Claude Opus 5