Anthropic ships Fable 5 — its most powerful public model yet

On 9 June 2026, Anthropic released Claude Fable 5 — the first generally-available model from its Mythos tier, the family that sits above Opus. Anthropic calls it its strongest public model for coding, knowledge work, vision and research. On some benchmarks it scored 10%+ above Opus 4.8 (per Anthropic — attributed, not asserted as universal). It can run autonomously for longer stretches than any prior Claude, making it more useful for multi-step agent tasks. A more capable, restricted version — Mythos 5 — sits above it in the tier structure, available only to vetted partners.

The catch. In cybersecurity, biology, chemistry and distillation, Fable 5 blocks responses and silently hands off to Opus 4.8. The most capable public model is also the most tightly fenced — a deliberate safety design, not a gap. Developers building in those domains need to account for the silent downgrade.

Pricing. Free on Pro, Max, Team and seat-based Enterprise plans through 22 June 2026. From 23 June: $10 per million input tokens, $50 per million output — roughly double Opus 4.8, per Anthropic. Subscription access returns 'when capacity is sufficient'. That's the language of a company rationing an expensive-to-run resource.

What's next. The open questions: whether the export-control directive becomes a permanent framework or a temporary block; whether Fable 5 returns to public access and on what terms; and whether the $10/$50 pricing survives competitive pressure from open-weight frontier alternatives now appearing monthly. Anthropic is also reportedly heading towards a large public offering, which makes a state-of-the-art launch strategically timed. The model's suspension is a complication, not a crisis.