Claude Opus 5 outperforms Claude Fable 5 while costing half as much

Anthropic has launched Claude Opus 5, a new AI that outperforms Claude Fable 5 in many areas while costing half as much. This new AI is also available to Claude Pro subscribers at no extra charge.

AI continues to evolve rapidly. Although Claude Fable 5, the consumer version of Mythos, is still relatively new, Anthropic has already unveiled a new AI called Claude Opus 5. In theory, Opus 5 sits in a lower category than Claude Fable 5. In practice, however, this new AI matches Fable 5 - and is even more capable for certain tasks. Opus 5’s key advantage is its lower price. It is included as standard for Claude Pro subscribers, whereas access to Fable 5 requires users to buy credits on top of their subscription fee.

Claude Opus 5: lower cost and leading performance

Despite delivering cutting-edge performance, Claude Opus 5 costs less than Claude Fable 5. More specifically, Anthropic says that Opus 5 is half the price. Yet, according to the assessments published by Anthropic, Claude Opus 5 outperforms every other AI model in certain areas, including office task automation. Opus 5 is now the market leader in the GDPval-AA v2 assessment, which measures how well models carry out real-world tasks. It also surpasses other AIs for computer use - and therefore agentic tasks - as well as for research.

According to Artificial Analysis’ intelligence index, Claude Opus 5 could even be the best AI on the market, ahead of Claude Fable 5, which the new model beats by one point in that index. In other words, Anthropic now offers an AI that excels at automating repetitive everyday tasks and is included with the Claude Pro subscription, unlike Fable 5.

An AI less likely to go off the rails

As well as being highly effective for productivity, Claude Opus 5 is also less dangerous. “As part of rigorous assessments conducted in collaboration with partners from the private and public sectors, we found that it remains below Mythos 5 in both biology research and offensive cybersecurity,” Anthropic states. The company also chose not to train this new AI on cyber tasks, meaning it can identify security flaws but remains far behind Claude Mythos when it comes to exploiting vulnerabilities.

Tests have also shown that Claude Opus 5 is more closely aligned with Claude’s “constitution” than Opus 4.8, Sonnet 5 or Fable 5. Anthropic further says that this new AI “has the lowest rates of deceptive behaviour” and is the least “likely to be tricked and misused.”

Claude Opus 5 safety benefits

In addition to its lower cost, which makes it more accessible, and its strong task-automation capabilities, Claude Opus 5 also offers a safety advantage.

What we think

With the launch of Claude Opus 5, Anthropic’s Claude subscription has suddenly become more competitive. Before this release, OpenAI could point to the inclusion of GPT-5.6 Sol in its subscription, while Anthropic required Claude Pro subscribers wanting to use Claude Fable 5 - then the most powerful consumer AI - to purchase usage credits.

Claude Opus 5 also makes Anthropic more competitive against Chinese AIs, which are becoming increasingly capable. Recently, the launch of Kimi K3, which is approaching the leading American AIs, shook up the entire industry.

The arrival of Google’s next flagship model is also eagerly anticipated. The Mountain View company recently said that it had already begun testing Gemini 3.6 Pro. Google is also developing Gemini 4, an entirely new generation of AI.