On Friday, Anthropic unveiled Claude Opus 5, a new AI model the company describes as approaching the frontier intelligence of its Fable 5 model at half the cost. According to Anthropic, Opus 5 surpasses Fable 5 in both programming tasks and knowledge-intensive applications.

Introducing Claude Opus 5. It's a thoughtful and proactive model that comes close to the frontier intelligence of Fable 5 at half the price. pic.twitter.com/GQWhcq2CQL

— Claude (@claudeai) July 24, 2026

Opus 5 is priced at $5 per million input tokens and $25 per million output tokens — a 50% reduction compared to Fable 5. Anthropic is marketing the model as one "designed to be used every day," positioning it as an economical option for organizations seeking to optimize their AI spending. The launch reflects a broader industry shift as leading AI labs increasingly compete not only on raw capability but on cost-per-task, a metric that has become central to enterprise procurement decisions as companies scale AI usage from pilot projects to production workloads.

Within Anthropic's product hierarchy, Opus 5 ranks beneath both Fable 5 and Mythos 5, bringing the Claude 5 family to four distinct models. Mythos 5 remains the company's flagship in terms of raw capability, but its availability is restricted to a select group of enterprise clients due to its sophisticated cybersecurity functionalities.

Economic Pressures Reshape Model Development

Corporate clients have grown more conservative in their artificial intelligence investments. Dianne Penn, Anthropic's head of product management for research, told CNBC that organizations prioritize quality over price alone.

"If it's a cheaper model or a cheaper offering, but it's not accomplishing a similar level of quality, it's actually not useful," Penn said.

Anthropic faces competitive pricing pressure from industry giants including Microsoft, Amazon, and Google, as well as from emerging Chinese AI ventures. OpenAI introduced its GPT-5.6 suite earlier this month, while Google is reportedly preparing a Gemini 3.5 Pro rollout, though reports indicate possible delays.

Last week, China's Moonshot AI debuted its open-weight Kimi K3 model. The research lab claims performance metrics nearly identical to both Fable 5 and GPT-5.6 Sol. The rapid emergence of open-weight alternatives has added further pressure on closed-model providers to demonstrate value through performance, safety infrastructure, and enterprise-grade features.

Public Listing Competition Intensifies Rivalry

These product launches come as the major AI companies chart their paths to public markets. Anthropic submitted confidential IPO registration documents in June and remains on track for a 2026 stock market debut.

OpenAI filed similar paperwork shortly after, though insiders suggest its public offering may be delayed until 2027. The company also faces legal action from Apple alleging misappropriation of confidential information, further complicating its IPO timeline. The convergence of multiple AI companies toward public listings is expected to subject the sector to greater financial scrutiny, with investors likely to weigh revenue growth against the substantial compute costs associated with training and serving large language models.

Earlier this year, the Trump administration mandated that Anthropic withdraw Mythos 5 and Fable 5, citing national security concerns. Market access resumed after approximately two weeks of regulatory discussions.

Anthropic continues to collaborate with federal agencies on pre-release security evaluations, including those conducted for Opus 5. The company confirmed that Opus 5 demonstrates reduced capabilities compared to Mythos 5 in offensive cyber operations and biological research applications. The voluntary security review process has become a differentiator among AI labs as policymakers weigh export controls and deployment restrictions on advanced models.

Unlike Fable 5, Opus 5 implements a zero-retention policy for user information. Anthropic continues to collect Fable 5 interaction data as part of its safety oversight protocols. The zero-retention option addresses growing enterprise demand for data governance controls, particularly in regulated industries such as finance and healthcare where compliance requirements restrict how vendor platforms handle sensitive inputs.

The AI development landscape remains intensely competitive as organizations navigate the intersection of capability, economics, and regulatory compliance during what is shaping up to be a pivotal year for the sector.