Anthropic Releases Claude Fable 5.1 With 45 Percent Lower Costs for AI Agents

Anthropic released two new AI models on September 2, 2026: Claude Fable 5.1 and Claude Mythos 5.1. The company says Fable 5.1 can cut the cost of highly agentic workloads by up to 45 percent compared to its predecessor.

Fable 5.1 is broadly available and designed for coding, knowledge work, and scientific research. Anthropic says it delivers similar or better performance than the previous generation at lower reasoning settings, making it more economical for businesses running sustained AI agent workflows.

Mythos 5.1 is more capable but remains restricted to vetted organizations. Anthropic says the model has advanced cybersecurity and biology capabilities that require additional safeguards before wider deployment.

Anthropic also reduced cache-read pricing and tightened safety filters, making it less likely that legitimate cybersecurity and scientific requests will be blocked. Fable 5.1 can now help identify software vulnerabilities in source code, though more sensitive exploit-generation capabilities remain reserved for restricted models.

The cost reduction is significant for enterprises that care about the economics of running AI agents continuously, not just how well a model performs on isolated benchmarks. Agentic AI systems that can take actions, use tools, and complete multi-step tasks are increasingly central to how businesses plan to use AI.

Anthropic secured a $35 billion cloud-computing deal with Lambda earlier this year to support the infrastructure needed for its growing model lineup.