What Claude Fable 5 Is and Why It Matters
Claude Fable 5 is Anthropic’s newest generally available reasoning AI model that offers Mythos-level performance in coding, knowledge work, and long-context tasks while applying extra safeguards on higher-risk uses such as cybersecurity and biology. Anthropic positions it as the most capable of its broadly accessible Anthropic AI models, matching the underlying Claude Mythos 5 system but with more restrictive behavior around sensitive topics. In practice, that means everyday users and teams can gain Claude Fable 5 access through the Claude API and product plans without entering a specialized security program. By routing risky prompts to the older Claude Opus 4.8 model, Anthropic aims to share advanced reasoning AI capabilities more widely while reducing the chance of misuse. This two-track design turns Fable 5 into the main Claude Mythos alternative for most people.
Two-Tier Strategy: Open Fable, Restricted Mythos
Anthropic now uses a clear two-tier model strategy: Claude Fable 5 for general access and Claude Mythos 5 for vetted, high-stakes users. Both draw on the same core model, but Mythos 5 has fewerbuilt-in limits and is reserved for selected cybersecurity, infrastructure, and future life sciences partners through programs like Project Glasswing. Those organizations can upgrade from Claude Mythos Preview to Mythos 5, gaining stronger offensive and defensive security capabilities that Anthropic considers too risky for public release. Meanwhile, Fable 5 stays broadly accessible as the flagship Claude Mythos alternative, with classifiers that detect offensive cybersecurity tasks, sensitive biology and chemistry, or attempts at model distillation. When those triggers fire, requests are redirected to Claude Opus 4.8, and Anthropic notes that more than 95 percent of Fable 5 sessions proceed without any fallback intervention.
Performance: From Massive Codebases to Complex Research
Anthropic is pitching Claude Fable 5 as a workhorse for demanding, multi-stage tasks that need long context, memory, and careful reasoning. The company says it outperforms all previous general Claude releases across software engineering, vision, scientific research, and knowledge work. In early testing, Stripe reported that Fable 5 completed a codebase-wide migration in a 50-million-line Ruby repository in one day, work they estimate would have taken a team more than two months manually. Anthropic also cites top performance on Cognition’s FrontierCode benchmark at medium effort, and strong senior-level reasoning on Hebbia’s Finance Benchmark. On the perception side, Fable 5 can extract numeric data from scientific figures, rebuild web app source code from screenshots, and has even completed Pokémon FireRed using a vision-only harness. Together, these results suggest that Fable 5’s reasoning AI capabilities are approaching the Mythos family’s headline performance.
Safety, Safeguards, and Responsible Access Controls
To address concerns about AI-enabled cyberattacks and bioterror, Anthropic has added strict safeguards around Claude Fable 5. For offensive cybersecurity, exploit development, and defense evasion, Anthropic reports that “Fable 5 complied with zero harmful single-turn requests,” even when subjected to 30 public jailbreak techniques. Biology and chemistry questions that might enable misuse are blocked at the Fable layer and routed to Claude Opus 4.8 instead. Internally, Anthropic runs classifiers to flag high-risk content, and in external testing it says more than 1,000 hours of bug bounty work produced no universal jailbreaks, while independent red-teamers could not break long-form agentic tasks. The company is also working with public authorities on plans for Claude Mythos 5 access, using trusted pathways like Project Glasswing and specialized biology programs to keep the most powerful capabilities on a tighter leash.
Access, Pricing, and API Paths for Developers
Claude Fable 5 access arrives through both Anthropic’s product plans and the Claude API, where it appears as claude-fable-5 for developers. During the initial rollout window, paying subscribers on Pro, Max, Team, and enterprise plans can try Fable 5 directly before it shifts to a usage-credit model after June 22, with continued access through pay-as-you-go credits from June 23. On the API side, both Fable 5 and Mythos 5 share the same listed pricing of USD 10 (approx. RM46) per million input tokens and USD 50 (approx. RM230) per million output tokens, which Anthropic says is less than half the cost of the earlier Claude Mythos Preview. This combination of flat per-token pricing and broad API availability means developers can integrate frontier-level reasoning into products and workflows without joining a restricted program, while Mythos 5 stays limited to trusted cybersecurity and critical infrastructure partners.
