Anthropic’s new Claude model comes close to Fable 5 at half the cost

Google just pulled the covers off of Gemini 3.6, as well as two new entries in the 3.5 Flash line. Not to be outdone, Anthropic is unveiling a new model of its own. This model adds on to the Opus series and is said to give the company’s powerful Fable 5 model a run for its money.

Anthropic has announced the launch of Claude Opus 5, which is described as being more efficient than other models. In the blog post, the company heaps praise upon the model, claiming that it approaches Fable 5 levels of intelligence, but at half the cost. It also claims that Opus 5 is “state-of-the-art” when it comes to coding and knowledge work, but trails behind Mythos 5 on cybersecurity tasks.

In terms of software engineer work, Anthropic states that Opus 5 doubles the performance of Opus 4.8 at a lower cost per task, according to Frontier-Bench v0.1. Similar results were reportedly seen when it came to knowledge work and problem-solving tasks. The model also improves upon scientific research, bettering its predecessor in every life sciences evaluation.

It’s worth pointing out that Opus 5 is said to output stronger visuals, is better at verifying and iterating on its work, and has improved on cyber tasks. Opus 5 is also said to come close to Mythos 5 in finding cybersecurity vulnerabilities. Additionally, Anthropic shares that Opus 5 is its most aligned model yet, adhering to Claude’s Constitution better than any other model.

Claude Opus 5 is available starting today on all platforms. It is priced at $5 per million input tokens and $25 per million output tokens. You’ll also have access to Fast Mode, which allows it to run 2.5 times the default speed. However, Fast Mode costs double the base price of Opus 5.