Anthropic has announced two new AI models: Claude Fable 5.1 and Mythos 5.1. Fable 5.1, in particular, promises better performance than the previous version. But the most important figure is the cost: 25% lower on average, and up to 45% lower for complex agentic tasks.
The reduction comes from pricing on data that has already been processed and stored in cache. As a result, automation pipelines that process large volumes of repeated context see the biggest savings. Anthropic has also responded to customer criticism about data retention and overly strict safeguards, making the model easier to use in real enterprise settings.
For Italian marketing managers evaluating AI agents for content automation, scalable customer service or digital workflow orchestration, this update significantly lowers the cost barrier to entry. At SHM Studio, we keep a close eye on these developments to identify the most practical applications for SMEs and the mid-market. In short, Claude Fable 5.1 is more than a technical update: it's a sign that agentic AI is becoming accessible to smaller business budgets.
What changed with Claude Fable 5.1 and Mythos 5.1
On September 1, 2026, Anthropic released two new models in the Claude family: Fable 5.1 and Mythos 5.1. The announcement, covered in detail by The Verge, is not just about performance. Anthropic explicitly said it listened to enterprise customers' criticism on three fronts: pricing, data retention and overly rigid safeguards.
Fable 5.1 keeps and improves on the capabilities of Fable 5. But the most significant change is structural: the new model costs 25% less on average. For complex agentic tasks, the savings reach 45%. This comes from lower pricing on data that has already been processed and stored in cache.
That means agentic architectures that reuse large contexts, typical of marketing automation, customer service or data enrichment workflows, benefit the most. This is not a marginal saving: for high-volume pipelines, the cost difference can be decisive in ROI assessments.
The number that matters: -45% on agentic tasks
It helps to distinguish between two types of use. The first is standard conversational use. The second is agentic use, where AI agents handle multi-step tasks with recurring contexts.
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.
