Claude Fable 5.1: What Anthropic’s New Model Changes for Coding, Research, and Costs

Anthropic has released Claude Fable 5.1 and Mythos 5.1, describing them as its most capable AI models yet. The two models share the same base model but differ in safety guardrails: Fable 5.1 is broadly available, while Mythos 5.1 is restricted to special access programs for cybersecurity and life sciences. The launch focuses on stronger coding performance, improved text quality, and better handling of long, tool-heavy tasks.

Anthropic says Fable 5.1 costs about 25 percent less than Fable 5 for typical workloads and can save up to 45 percent on complex agentic workflows with many tool calls. The key pricing change is a cache-read cut from $1 to $0.25 per million tokens, while other API prices remain $10 per million input tokens and $50 per million output tokens. An update in the article notes that Artificial Analysis disputes the broad savings claim for max-effort runs, saying Fable 5.1 can cost 20 percent more per task than Fable 5 because it uses roughly 1.7 times as many output tokens.

Fable 5.1 posts notable gains on agentic benchmarks. On Terminal-Bench-Science 0.1, it scores 52.6 percent, more than double Fable 5’s 24.7 percent and ahead of GPT-5.6 Sol at 22.4 percent. On Terminal-Bench 4.0 for agentic coding, Fable 5.1 reaches 55.8 percent, while Mythos 5.1 reaches 60.9 percent; the article notes that real-world translation of these gains will become clearer over time.

Claude Fable 5.1 is available immediately across Anthropic’s platforms, including AWS, Google Cloud, and Microsoft Azure, and developers can use it through the API as claude-fable-5-1. Claude Mythos 5.1 is limited to US organizations through the Cyber Verification Program and the Life Sciences Verification Program, with plans to expand access to international partners. The 5.1 models are also the first Claude models to ship with built-in watermarks, and Anthropic is launching a detection API in private preview for groups such as regulators, media outlets, fact-checkers, and research institutions.

The article says the new models use less aggressive safety filters for cybersecurity, biology, and medical questions compared with earlier versions. Cybersecurity false positives are reported as 60 percent lower, while biology-related filters fire 85 percent less often on harmless basic biology and medicine questions. Fable 5.1 can now identify software vulnerabilities, but penetration testing and exploit generation are still routed to Opus models; Anthropic is also limiting a documented distillation technique by changing how new API accounts handle prior context and thinking transcripts.