Claude Opus 5 targets Fable performance at half the price

Anthropic has released Claude Opus 5, positioning its latest flagship model as a faster, more efficient option that approaches the performance of Claude Fable 5 at half the price.

The model targets coding, knowledge work, and agent workflows. Anthropic says it delivers stronger performance on complex coding tasks, improved autonomy, better error recovery, and more frequent checks of its own work. It is also designed to require less back-and-forth from users.

Cristian Rivera, a software engineer at Stripe, described Opus 5 as a “clear generational step up from Opus 4.8.”

“Over one weekend, I gave it a chief-of-staff role over my dev environments: it built its own monitor, drove each box, and pulled me in only for the judgment calls.”

Performance and token efficiency

Anthropic says Opus 5 has the strongest knowledge-work performance of any model it has released. The company also claims that it comes close to Fable 5's capabilities at half the price, while providing substantial gains in complex software development.

Cursor co-founder Sualeh Asif said the model performed just below Fable 5 on CursorBench and displayed many of the same behaviors.

Recommended reading

Google Zero is forcing websites to rethink the web

“Claude Opus 5 delivers near Fable 5 intelligence at Opus speed and cost. On CursorBench, it’s just under Fable 5 and has many of the same behaviors.”

Harvey, which develops AI legal tools, reported improvements in legal-agent work, particularly in corporate governance and arbitration. Its head of applied research, Niko Grupen, said Opus 5 achieved similar performance while generating 26% fewer tokens on average than Opus 4.8 at maximum reasoning.

Lower token use can reduce costs and allow an agent to complete more work within a session. Anthropic also says Opus 5 “overperforms on biology tasks compared to Opus 4.8,” though it has not defined that term in detail.

Pricing and availability

Anthropic says Opus 5 uses the same API pricing as Opus 4.8:

• $5 per million input tokens
• $25 per million output tokens

For Claude Max subscribers, Opus 5 becomes the default model with standard limits on all paid plans. It is available now to every user tier that previously had access to Opus 4.8. Anthropic continues to recommend Fable 5 for its most ambitious, days-long autonomous projects.

The company has also introduced Fast mode for Opus 5 as a research preview. It offers full Opus 5 capability with 150% faster token generation at twice the standard price. Claude Code Max subscribers can purchase the feature with additional usage credits.

Other changes include automatic API fallbacks when a safety classifier declines a request, plus the ability to change external tools during a conversation without clearing the prompt cache. Developers can therefore disable tools when they are no longer needed, supporting cost control and security.

Anthropic expects Opus 5 to become a daily-use model, particularly for enterprise customers.