Anthropic dropped two new Claude models this week — Fable 5.1 for general release and Mythos 5.1 for restricted cybersecurity and life sciences work — and the benchmark numbers are genuinely impressive. On Terminal-Bench-Science 0.1, Fable 5.1 hits 52.6 percent, more than double its predecessor’s 24.7 percent and far ahead of GPT-5.6 Sol’s 22.4 percent. Agentic coding tells a similar story: Fable 5.1 scores 55.8 percent on Terminal-Bench 4.0, while Mythos 5.1 reaches 60.9 percent, compared to Fable 5’s 42.0 percent. Those are not incremental gains. Anthropic researcher Felix Rieseberg also confirmed the model dials back its tendency to overuse bullet points and bold text, making chat responses sound more natural — a quality-of-life fix that anyone who’s used Claude extensively will appreciate.
But the pricing story is messier than Anthropic’s announcement suggests. The headline claim is that Fable 5.1 costs up to 45 percent less than Fable 5, driven largely by cache reads dropping from $1 to $0.25 per million tokens. That’s real for workloads that lean on caching. Yet Artificial Analysis, which helped with pre-release testing, ran the numbers and found that at max effort, Fable 5.1 actually costs about 20 percent more per task than Fable 5 — $3.76 versus $3.14 — because it burns through roughly 1.7 times as many output tokens. Even more damning: Opus 5 delivers a score only three points lower on the Intelligence Index at $2.34 per task. For enterprise buyers who already complained about Fable 5’s pricing, that math matters.
The safety filtering also got a tune-up that should reduce friction. Cybersecurity filters now trigger 60 percent fewer false positives, and biology-related queries see 85 percent fewer misfires on harmless questions. For the first time, Fable 5.1 can identify software vulnerabilities, though penetration testing and exploit generation still route to Opus models. Both 5.1 models also ship with built-in watermarks, with a detection API in private preview for regulators, media, and researchers.
Here’s the tension: Anthropic has been under pricing pressure since Opus 5 launched in late July, matching or beating Fable 5 on most benchmarks at half the cost. Fable 5.1 closes the capability gap, but the cost-efficiency picture depends heavily on which effort level you choose and whether your workload benefits from cached reads. The real test over the coming weeks is whether enterprise adoption picks up or the Opus line keeps eating Fable’s lunch on value.
