Stock Markets July 24, 2026 01:02 PM

Anthropic Debuts Opus 5, Promising Fable 5-Like Performance at Lower Cost

New Claude-model iteration aims for workplace and programming use with efficiency and adjusted safety settings

By Jordan Park
Share
Twitter Reddit Facebook LinkedIn

Anthropic unveiled Opus 5, a new Claude-family AI model that the company says approaches the capabilities of Fable 5 while costing roughly half as much. The firm positions Opus 5 for routine office work and coding, emphasizing improved efficiency over the prior Opus 4.8 release and altered safety constraints tied to its comparative limitations versus Fable 5.

Anthropic Debuts Opus 5, Promising Fable 5-Like Performance at Lower Cost
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • Anthropic launched Opus 5, a Claude-family model the company says nears Fable 5’s capabilities while costing about half as much - impacts AI and enterprise software purchasing decisions.
  • Opus 5 is positioned for daily office work and programming, and the company asserts it is more efficient than the prior Opus 4.8 release - relevant to cloud compute and developer tool adoption.
  • Opus 5 carries less restrictive safeguards than Fable 5 based on testing that found it less capable of exploiting cyber vulnerabilities; Fable 5 remains the choice for extended, autonomous projects.

SAN FRANCISCO, July 24 - Anthropic on Friday introduced Opus 5, the latest addition to its Claude lineup. The company said the model approaches the performance of its more capable sibling, Fable 5, while being offered at about half the price.

Anthropic described Opus 5 as particularly well suited to everyday office tasks and computer programming assignments. In an interview, product leader Dianne Penn said the model’s launch represented another step in rapid development, noting that Opus 5 is more efficient than the May release, Opus 4.8.

"We’re building and continue to consistently deliver frontier intelligence and bring that as accessibly as possible with every model generation," Penn said.

According to Anthropic’s internal testing, Opus 5 is less capable than the company’s top-tier AI at exploiting cyber vulnerabilities. That relative limitation has informed the company’s approach to safety restrictions: Opus 5 carries less restrictive safeguards than Fable 5 because it showed a lower ability to perform such harmful tasks in tests.

Anthropic also reported that Opus 5 proved harder to trick into misuse compared with its other currently deployed models.

Fable 5, released in June, had a temporary interruption in availability after U.S. authorities raised concerns that its capabilities might be repurposed for foreign military intelligence. Penn advised users to select models based on task scope: Opus 5 is the value choice for routine work, whereas Fable 5 is more appropriate for "days-long, very autonomous projects," she said.

When asked about the Kimi K3 "open" model produced by China-based Moonshot - which U.S. officials have accused of drawing on Anthropic’s work without permission - Penn said it "remains to be seen" how open-weight models broadly will fare when deployed on the complex, real-world projects that customers currently ask Claude to handle.

Anthropic explained that open-weight models differ from proprietary models because they allow users to download, run and customise the underlying model weights - effectively enabling external parties to run and modify the model’s virtual 'brain' locally.


Implications

The company frames Opus 5 as an efficiency-focused alternative intended for frequent, everyday tasks and for users seeking lower-cost compute. At the same time, Anthropic draws a distinction between that value orientation and the higher-capability, more restricted Fable 5 for extended autonomous workloads.

Risks

  • National security and regulatory scrutiny - Fable 5 was temporarily unavailable after U.S. concerns that its capabilities could be diverted to foreign military intelligence, which could affect availability for high-capability models; this risk touches government, defense, and high-security enterprise use cases.
  • Cybersecurity considerations - differences in models’ ability to exploit vulnerabilities require tailored safety measures and may influence enterprise deployment choices, affecting sectors that prioritise secure AI integration.
  • Performance uncertainty for open-weight models - Anthropic highlighted uncertainty about how open-weight offerings like Moonshot’s Kimi K3 will perform on complex real-world tasks, a factor that could influence enterprises and developers evaluating model portability and customisation.

More from Stock Markets

Beumer Group Takes EU to Court Over Vanderlande-Siemens Airport Logistics Deal Jul 24, 2026 Aerospace Struggles to Replace Retiring Engineers as AI Pulls Talent Elsewhere Jul 24, 2026 White House to Open Trade Probe After EU Imposes New Fine on Google Jul 24, 2026 Qualcomm to Apply Double-Digit Price Hikes for Shipments After Sept. 1 Jul 24, 2026 Anthropic debuts Claude Opus 5 as lower-cost option for workplace AI tasks Jul 24, 2026