Claude Opus 5 Outscores Fable 5 on Most Benchmarks—At Half the Price

Claude Opus 5 does out today. It’s cheaper for businesses to use than Anthropic’s flagship model, Claude Fable 5, which the company had positioned as the everyday frontier product for paying users. Moreover, Opus 5 also performs better on important benchmarks.

To understand where Opus 5 fits, Anthropic’s lineup consists of four levels. Haiku is fast and cheap. Sonnet is middle class. Opus is the heavy workhorse. On top of that is the Mythos class – a tier Anthropic introduced this spring – which includes Claude Fable 5 for the public, and Claude Mythos 5, a less-restricted version reserved through Project Glasswing for vetted cybersecurity researchers and critical infrastructure operators.

Fable 5 has struggled as a flagship for subscribers. It launched on June 9, was withdrawn globally three days later after the US government issued an emergency export control order citing a jailbreak vulnerability, and returned on June 30 – only to immediately switch to a credits-only model, which is no longer included in standard plans. Opus 5 now fills the gap that Fable 5 could not fill.

Lovable, a developer platform with millions of users, ran Opus 5 based on internal evaluations and noted that the gains extend beyond the raw scores: “It’s not only better on our toughest coding tasks, up 22% over Opus 4.7, it’s more stable, with much less variance,” Fabian Hedin said in a statement shared by Anthropic.

The benchmarks

It may sound strange, but Opus beats Fable on almost everything important to the everyday user, while not being labeled as Mythos-class like Fable.

On Frontier-Bench v0.1 – a benchmark that tests whether AI coding agents can complete real software engineering tasks end-to-end, scored as a percentage of successful tasks – Opus 5 achieved a score of 43.3%. Myth 5 came to 33.7%. OpenAI’s GPT-5.6 Sol, Anthropic’s main commercial rival, scored 34.4%.

The largest margin is on ARC-AGI-3, a real-world problem-solving test built around new puzzles that a model could not have remembered from training data, scored as a percentage of puzzles solved. Opus 5 reached 30.2%; GPT-5.6 Sol scored 7.8%; and Fable 5 has not been tested at all. On GDPval-AA v2 – a knowledge work benchmark scored via Elo ratings, the chess-style ranking system used to measure relative performance on real-world professional tasks – Opus 5 reached 1,861, versus Fable 5’s 1,747 and GPT-5.6 Sol’s 1,736.

Zapier tested Opus 5 on AutomationBench, an evaluation that assesses whether a model can execute an entire business workflow from start to finish without human assistance. Their verdict: The model “took a raw workbook on account health and performed a full suite of churn prevention: flagging risky accounts, alerting the appropriate owner, and summarizing for retention operations. Previous models have not succeeded; Opus 5 reached 100%.”

Anthropic is also pitching Opus 5 as a research upgrade. Ultima Genomics, a DNA sequencing company, said the model “behaves more like a careful scientist than any model we’ve run. It looks for the right statistical tests to rule out confounders, compares its own results to independent methods, and stays on track through long multi-step analyses.”

That said, these two areas – legal and healthcare – are the only ones where Fable 5 excels by a small margin.

The release comes a week after Moonshot AI, a Beijing-based startup backed by Alibaba, unveiled Kimi K3: an open-weight model with 2.8 trillion parameters (meaning anyone can download the underlying code to run it independently) that Moonshot describes as the world’s largest open AI system. Independent benchmarks consistently place Kimi K3 third overall, behind both Fable 5 and GPT-5.6 Sol, beating the two in specific areas.

Opus 5 is now available via API for $5 per million input tokens and $25 per million output. (Tokens are the basic unit of information that an AI model can process in both input and output). There is also a fast mode available that has approximately 2.5 times the standard speed, at twice the base price: $10 per million input tokens and $50 per million output.

This release may put an end to concerns about Fable 5’s lack of public availability. Opus is also available to everyone via subscription.