Anthropic is responding to the loudest complaints about its Claude models — price, privacy, and paternalistic safety filters — with a pair of updates that suggest the company is finally listening to enterprise customers. Fable 5.1 and Mythos 5.1 arrive with a headline number that matters for anyone running agentic workloads at scale: up to 45 percent lower costs on complex agentic tasks, and roughly 25 percent cheaper in typical usage. The savings come largely from reduced pricing on cached data, which is exactly where agentic systems rack up bills since they’re constantly reprocessing context.
The early reception from people who actually build on this stuff is notably warm. Dan Shipper, CEO of Every, says Fable 5.1 is “the strongest coding model we’ve used” while being fast and token-efficient — and he adds a dig that hints at a real pain point: it “actually speaks like a normal person.” Box CEO Aaron Levie claims his company’s agent caught subtleties in data that Fable 5 missed on the same test. Those aren’t benchmark graphs; they’re production anecdotes from paying customers, which carry more weight than any leaderboard. Meanwhile, a user going by Lisan al Gaib notes that Mythos 5.1 on low reasoning settings scores the same as the previous model cranked to Max reasoning — efficiency gains that could quietly reshape how people configure these models.
On the safety front, Anthropic is making a meaningful distinction. Fable 5.1 gets “more precise safeguards” that won’t block basic biology questions the way its predecessor did, and it’s now allowed to identify software vulnerabilities. But Mythos 5.1 keeps the same biology restrictions, and certain cybersecurity tasks like penetration testing and exploit generation still get redirected to Opus models. That split personality — looser for Fable, tighter for Mythos — reflects Anthropic’s attempt to serve different markets without fully abandoning its cautious brand.
The privacy story is where things get genuinely interesting for regulated industries. Anthropic’s Enterprise Frontier Safeguards promise “complete privacy” by storing data on the customer’s own cloud servers rather than Anthropic’s infrastructure, rolling out later this fall. That’s a direct shot at the data residency anxieties that have kept financial and healthcare companies on the sidelines. Fable 5.1 is available everywhere now; Mythos 5.1 remains gated behind Project Glasswing. The price drop paired with data residency could force OpenAI and Google to answer the same questions Anthropic just did — and neither has been eager to discuss data retention specifics.
