Anthropic Launches Claude Opus 5 With Near-Fable Intelligence at Half the Cost

Anthropic has officially launched Claude Opus 5, a new flagship model that delivers intelligence close to the Claude Fable 5 frontier AI model at half the price. Available today on all platforms, the model becomes the default for Claude Max subscribers and the strongest model on Claude Pro.



Opus 5 arrives just two months after Claude Opus 4.8 and roughly six weeks after the debut — and later restoration — of Anthropic's more capable but heavily restricted Fable 5 and Mythos 5 models. Anthropic says Opus 5 is "much stronger at verifying its work and iterating carefully until it succeeds," citing a benchmark in which the model built its own computer vision pipeline from raw pixels to reconstruct a 3D machine part after receiving an intentionally incomplete prompt. No competing model completed the task under the same constraints.


The company says Opus 5 establishes new state-of-the-art results on coding and knowledge work evaluations such as Frontier-Bench and GDPval-AA. On CursorBench 3.2, Opus 5 performs within 0.5 percent of Fable 5's peak score at half the cost, while on OSWorld 2.0 it outperforms every other model at a given cost, surpassing Fable 5's best result at just over one-third the cost. Anthropic also highlights stronger performance for scientific research, with improvements across life sciences, structural biology, organic chemistry, and bioinformatics evaluations.



For developers, Opus 5 keeps the same pricing as Opus 4.8 at $5 per million input tokens and $25 per million output tokens for everyday coding, knowledge work, and agentic tasks. A Fast mode runs at roughly 2.5 times the standard speed for twice the base price. Alongside the model, Anthropic introduced two new beta features: mid-conversation tool changes, which let developers change which tools Claude can access without invalidating the prompt cache, and Automatic Fallbacks, which automatically route requests flagged by Opus 5's safety classifiers to the best available model instead of blocking the request.

Opus 5 uses cybersecurity safeguards similar to Opus 4.8, but with significantly lighter restrictions than Fable 5. Anthropic expects its safety classifiers to intervene roughly 85 percent less often than they do for Fable 5. The model is still blocked from binary-based vulnerability scanning, penetration testing, and exploit generation, but it can identify vulnerabilities in source code for defensive purposes. Unlike Fable and Mythos, Opus 5 does not carry a 30-day data retention requirement for general access. Anthropic also says it is the company's most aligned model to date, showing better adherence to Claude's Constitution, lower rates of deceptive behavior, and greater resistance to misuse than previous Claude models.