Claude Opus 5: Anthropic Challenges Fable at a Lower Price

Anthropic Unveils Claude Opus 5 with Optimized Performance

The company Anthropic has recently introduced the latest version of its artificial intelligence model, Claude Opus 5. This model is specifically designed to meet the needs of developers and businesses, focusing on significant improvements in coding, reasoning efficiency, and compatibility with prompt cache management tools. Opus 5 stands out with performance levels approaching those of the Fable model, while being offered at a reduced cost.

Highlights of Claude Opus 5

The Opus 5 model primarily focuses on enhancing coding processes, optimizing knowledge work, and managing agent workflows. According to Anthropic, Opus 5 uses fewer tokens while maintaining equivalent output quality. Developers can thus benefit from faster production and smarter tool control.

Expected Launch Following Mythos and Fable

The launch of Claude Opus 5 is not a surprise, especially after the recent releases of the Mythos 5 and Fable 5 models last month, followed by Sonnet 5 in early July. The transition of Opus from version 4.8 to 5 was inevitable. Opus 4.8 had already proven its effectiveness in managing complex coding projects, and early feedback suggests that Claude Code could see a significant boost with this new version.

Testimonials from Satisfied Users

Cristian Rivera, a software engineer at Stripe, shared his experience: "Opus 5 is clearly a generational leap over Opus 4.8. In one weekend, it took over my development environments, building its own monitor and managing each box, only asking me for crucial decisions."

On his part, Niko Grupen, head of applied research at Harvey, stated: "Claude Opus 5 represents a notable advancement for legal agent work, surpassing previous models in areas like corporate governance and arbitration." He also noted that the model generates on average 26% fewer tokens while maintaining high reasoning quality.

Performance Comparable to Fable 5

Anthropic claims that Opus 5 approaches the capabilities of the Fable 5 model in many areas, but at half the cost. The company emphasizes that this model offers the best performance in knowledge work among all its released models, with substantial gains in complex coding. These improvements are directly beneficial for users of Claude Cowork and Claude Code.

The AI provider Cursor confirms this assessment, indicating that Opus 5 is poised to compete with Fable 5 in terms of capabilities. Sualeh Asif, co-founder of Cursor, stated: "Claude Opus 5 offers intelligence close to that of Fable 5, but at the speed and cost of Opus. On CursorBench, it is just below Fable 5 and exhibits many similar behaviors."

Increased Autonomy and Efficiency

Opus 5 is designed to be the flagship model for enterprises. Having used Opus 4.8 extensively with a Claude Max plan at $100/month, I encountered no limitations. In recent tests, I reached only 31% of my weekly usage limit. I plan to switch to Opus 5, which should allow me to accomplish even more complex tasks.

While Opus 5 excels in knowledge work, it also offers increased autonomy, reducing back-and-forth interactions, improving error recovery, and checking its work more frequently. Anthropic also highlights that Opus 5 outperforms its predecessor in biology tasks and other scientific research.

Pricing Maintained and Recommendations

Anthropic has decided to maintain the same pricing as for Opus 4.8. For token API users, the cost remains at $5 per million input tokens and $25 per million output tokens. For Claude Max subscribers, Opus 5 will be the default model with "standard limits on all paid plans."

Anthropic continues to recommend Fable 5 for the most ambitious and autonomous projects that require several days of work. I plan to integrate in-app purchases for my iPhone and Mac applications, and I am considering entrusting this project to Fable to assess its capabilities.

New Features and Announcements

Anthropic has also announced several new features:

• Fast Mode for Opus 5 on the Claude platform, offering token generation 150% faster at double the standard price. This feature is available for Claude Code subscribers with additional credits.

• Automatic Feedback: When a security classifier denies a request, the API automatically switches to another model to provide a helpful response.

• Tool Switching During Conversations: Developers can now swap external tools used by Claude without clearing the prompt cache, helping to control costs and ensure security.

Opus 5 is now accessible to all users who previously had access to Opus 4.8.