Anthropic today released Claude Fable 5, the first publicly available Mythos-class model, priced at $10/$50 per million tokens — less than half the Mythos Preview rate — and bundled free into paid plans through June 22.
Anthropic Releases Claude Fable 5 — Its First Public Mythos-Class Model
By Hector Herrera | June 9, 2026
Anthropic today released Claude Fable 5, the first publicly available model from its Mythos-class tier — the capability level above the Opus line that had been locked behind a restricted partner program since earlier this year. The release cuts the access price by more than half and bundles the model into existing paid plans through June 22.
Background
Mythos-class models represent Anthropic's highest-capability tier. Until today, access was limited to a small group of enterprise partners through Project Glasswing, an invitation-only early-access program. Most API developers and even paid subscribers had no way to reach Mythos-level outputs — they topped out at Claude Opus 4.8. Today's release changes that.
Get this in your inbox.
Daily AI intelligence. Free. No spam.
What Anthropic Released
- Model: Claude Fable 5, the first publicly available Mythos-class model
- API pricing: $10 per million input tokens / $50 per million output tokens — less than half the rate charged under Mythos Preview access
- Plan inclusion: Available at no additional charge to Pro, Max, Team, and Enterprise subscribers through June 22, 2026
- Safety architecture: Hard limits in three high-risk domains — cybersecurity, biology, and chemistry — with automatic fallback to Claude Opus 4.8 when those limits are triggered
- Release date: June 9, 2026
What the Safety Fallback Actually Means
The fallback mechanism is one of the more interesting engineering decisions here. Rather than returning an error or a refusal when a query touches the restricted domains, Fable 5 silently routes to Opus 4.8 and returns a lower-capability response. For most production applications, that's invisible. For developers building specifically around Mythos-tier outputs in sensitive domains, it's something to account for in testing.
This approach is consistent with how Anthropic has handled capability restrictions before — the goal is to keep applications running while limiting the highest-risk outputs, rather than breaking workflows with hard blocks.
What This Means for Enterprise Customers
The pricing tells the real story. At $10/$50 per million tokens, Fable 5 lands well below where enterprise buyers expected Mythos access to sit. Combined with the two-week free trial window for existing plan subscribers, Anthropic is clearly trying to generate adoption data quickly before the full pricing regime kicks in after June 22.
For teams currently on Opus 4.8, the decision calculus is straightforward: test Fable 5 before June 22, benchmark the output quality improvement against your specific use case, and decide whether the premium is worth it at full price. The free window exists for exactly that purpose.
What to Watch
The June 22 cutoff will be the first real signal of enterprise appetite for Mythos-tier pricing. Watch for capability benchmarks from the developer community over the next two weeks, and for Anthropic to publish any comparative performance data. If Project Glasswing partners have been quietly shipping Mythos-class products, their results will surface quickly now that the model is public.
Did this help you understand AI better?
Your feedback helps us write more useful content.
Get tomorrow's AI briefing
Join readers who start their day with NexChron. Free, daily, no spam.