claude-mythos-5 — pricing, context window and limits
| Status | current |
|---|---|
| Input / MTok | $10 |
| Output / MTok | $50 |
| Context window | 1,000K tokens |
| Max output | 128K tokens |
| Min cacheable prefix | 512 tokens |
| Effort levels | low, medium, high, xhigh, max |
| Thinking | always-on |
| Platforms | Claude API (Project Glasswing participants only) |
The same model, a different door in
Mythos 5 is not a separate model in any capability, pricing, or behavioural sense — it shares Fable 5's exact limits, pricing, and API behaviour in full. The only real difference is how you reach it: Mythos 5 is available exclusively through a specific participant access programme, while Fable 5 is the same underlying capability reachable through ordinary API access.
Why this entry exists separately at all
It would be easy to assume a second model id sharing identical specs is a mistake or a rounding error in a reference table, which is exactly why it's worth stating plainly: this is a deliberate, distinct id, not a duplicate to collapse. Code that might route to either — through a fallback, a feature flag, or an account-specific access grant — needs to treat both ids as equivalent for pricing and behaviour purposes, without assuming one is a fallback or a downgrade of the other.
What to actually do if you don't have program access
If your organisation isn't part of the access programme this id is gated behind, Fable 5 is the model to use — there is no capability, pricing, or behavioural reason to prefer one over the other beyond which access path is actually open to you.
Related
See claude-fable-5 for the full specification this model shares in every respect except its access path.
Verified 2026-08-08 against claude-api skill — shared/models.md + SKILL.md model table (canonical: https://platform.claude.com/docs/en/about-claude/models/overview, https://platform.claude.com/docs/en/pricing).