Related angle — everyday frontier tier. On July 24, 2026, Anthropic released Claude Opus 5: near Claude Fable 5 intelligence at half Fable’s listed price ($5/$25 vs $10/$50 per million input/output tokens). For the twin-model launch itself, see the series opener; this piece is the Opus slot only.
Confirmed product facts
- Available on Anthropic’s platforms; API id claude-opus-5.
- New default on Claude Max; strongest model on Claude Pro.
- Anthropic highlights coding and knowledge-work gains (e.g., Frontier-Bench, CursorBench claims) and cost/performance curves versus prior Opus and versus Fable.
- Safeguards: cyber classifiers less restrictive than Fable 5 in some areas (allows source-code vulnerability finding; blocks binary-based scanning, penetration testing, exploit generation per Anthropic). Biology safeguards similar to Opus 4.8; Fable biology blocks can route to Opus 5.
- Anthropic says Opus 5 remains behind Mythos 5 on cybersecurity exploitation and on the strongest biology research tasks.
- Fast mode (~2.5× speed) at 2× base price on the Claude Platform / usage credits in Claude Code.
Where it sits in the stack
Opus 5 is the everyday high-capability tier after Fable/Mythos 5: closer to frontier coding/agentic quality without Mythos gating or Fable’s full price. It does not replace Glasswing’s Mythos track.
Our take
In our view, Opus 5 is Anthropic’s pressure-release valve: keep Mythos gated, keep Fable expensive/safeguarded, and give developers a near-frontier Opus default that still trails Mythos on the sharpest dual-use edges.
Series: Anthropic Frontier · consequence file.
