Claude Fable 5

Anthropic · Released June 9, 2026

What It Actually Is

If Opus 4.8 was the promotion, Fable 5 is the corner office. Anthropic’s naming shift from musical tiers (Haiku, Sonnet, Opus) to literary ones (Fable, Mythos) isn’t just branding — it signals a new class of model. Fable 5 runs on the same Mythos-class architecture that powers the restricted Mythos 5, but with safety classifiers that make it safe for general use. Think of it as a supercar with the speed limiter set — still the fastest thing on the road, just with guardrails on certain turns.

The numbers tell the story. SWE-Bench Pro 80.3% doesn’t just beat GPT-5.5 (58.6%) — it embarrasses the entire field. FrontierCode Diamond at 29.3% means Fable 5 writes production-quality code five times more efficiently than GPT-5.5 (5.7%). On Hebbia’s Finance Benchmark — senior-level document reasoning, chart reading, root-cause analysis — it’s #1. On CursorBench, it opened up “a class of long-horizon problems that were out of reach for earlier models.”

But the most telling demonstrations aren’t benchmarks. Stripe migrated a 50-million-line Ruby codebase in one day — work that would have taken a full team two months. The model completed Pokémon FireRed using only raw screenshots — no maps, no helper tools, no game-state data. And when given persistent file-based memory playing Slay the Spire, its performance improved 3× more than Opus 4.8’s.

The safety story is worth understanding. Queries touching cybersecurity, biology, chemistry, or model distillation get automatically routed to Opus 4.8 — still a top-tier model, but not the full Mythos architecture. This happens in less than 5% of sessions, and Anthropic acknowledges some false positives on harmless queries. It’s the price of releasing a model this capable quickly and safely. The unrestricted Mythos 5 is reserved for vetted partners through Project Glasswing — where it’s already helping defend critical software infrastructure.

The real question is whether the price is worth it. At $10/$50 per million tokens, Fable 5 costs roughly 2× what Opus 4.8 does. But token efficiency partially offsets this — achieving FrontierCode-leading results at medium effort means less compute per task. For professionals whose time is worth more than their API bill, the math is simple. For everyone else, Opus 4.8 remains excellent. But if you want the best generally available AI model on the planet — the one where the gap widens as the task gets harder — this is it.

Key Strengths

• Mythos-class capability for everyone: Same underlying architecture as the restricted Mythos 5, but with safety classifiers that make it broadly available. Fable 5 is state-of-the-art on nearly all tested benchmarks — and the gap over competitors grows as tasks get more complex. This isn’t incremental; it’s a generational leap.
• Autonomous agent that actually delivers: Stripe compressed months of engineering into days — migrating a 50-million-line Ruby codebase in one day. The model plans, delegates to sub-agents, self-verifies with its own tests, and keeps going until the job is done. Multi-day autonomous sessions are the new normal.
• Vision breakthrough: State-of-the-art on vision tasks. Can extract precise numbers from scientific figures, rebuild web apps from screenshots alone, and complete Pokémon FireRed with vision only — no helper harnesses, no game state data. Earlier models needed complex scaffolding; Fable 5 needs eyes.
• Memory across millions of tokens: Persistent file-based memory improved its Slay the Spire performance 3× more than Opus 4.8. The model stays focused across million-token sessions and actually improves its outputs using its own notes. Long-context isn’t just a spec — it’s a working feature.
• Token efficiency wins the math: Despite 2× per-token pricing vs Opus 4.8, Fable 5 scores highest on FrontierCode even at medium effort. More work done per token means real-world cost per task is often competitive. The expensive model that saves money on hard problems.

Honest Limitations

• ⚠️ Access suspended for non-US nationals: On June 12, 2026, the US government issued an export control directive suspending all access to Fable 5 and Mythos 5 for any foreign national — whether inside or outside the United States. Anthropic has had to disable the model for all customers to ensure compliance. All other Anthropic models remain available. Anthropic disagrees with the directive and is working to restore access. Check their announcement for the latest status.
• Premium pricing is real: $10 per million input tokens, $50 per million output tokens — roughly 2× Opus 4.8 rates. Pro subscribers get included access through June 22, then usage credits kick in. Power users will feel the bill.
• Conservative safety routing: Safeguards trigger in <5% of sessions, routing flagged queries to Opus 4.8 instead. Some false positives on legitimate professional work (cybersecurity research, chemistry, biology). The guardrails reflect the dual-use power of the underlying model.
• Not the full Mythos 5: The unrestricted version is locked behind Project Glasswing for vetted cyberdefenders and researchers. What you get is explicitly a guarded version — extremely capable, but with training wheels on certain topics.
• Independent benchmarks pending: Launch-day claims are detailed and example-rich, but full LMSYS Arena, Artificial Analysis, and updated SWE-Bench third-party results are still emerging. Verify before you crown.