claude-mythos-5 — pricing, context window and limits
Fable 5's Project Glasswing counterpart, identical in pricing, limits and API behaviour apart from the model id. It ran no safety classifiers, so a refusal stop reason does not occur on it. Mythos 5.1 is its successor.
Price per million tokens
| Input | $10 |
|---|---|
| Output | $50 |
| Cache write, 5-minute TTL | $12.50 |
| Cache write, 1-hour TTL | $20 |
| Cache read | $1 (0.1x input) |
| Batch input / output | $5 / $25 |
Limits and behaviour
| Status | Current |
|---|---|
| Context window | 1M tokens |
| Max output | 128K tokens |
| Min cacheable prefix | 512 tokens |
| Effort levels | low, medium, high, xhigh, max |
| Effort default | high |
| Thinking | Always on — cannot be disabled |
| Forced tool_choice | Accepted |
| Platforms | Claude API |
| Access | Project Glasswing participants only (invitation) |
Lifecycle
| API model name | claude-mythos-5 |
|---|---|
| Retires no sooner than | 9 Jun 2027 |
| Successor | Claude Mythos 5.1 |
The same model as Fable 5, behind a different door
Mythos 5 shares Fable 5's limits, pricing and API behaviour in full. The difference is access: Mythos 5 is offered only to organisations admitted to Project Glasswing, while Fable 5 is the same capability through ordinary API access. Code that can route to either — through a fallback, a feature flag or an account-specific grant — should treat the two ids as equivalent for pricing and behaviour, not as an upgrade and a downgrade of each other.
The detail that makes it different from its successor
Mythos 5 ran no safety classifiers, so a response from it never ends with a refusal stop reason. Its successor does not share that property. Unlike Mythos 5, its successor runs safeguards, and which ones depends on the access programme an organisation was approved under — so Mythos 5 code that never checked for a refusal needs that check before it moves. The replacement is otherwise the same model as Fable 5.1, cheaper cache reads included.
If you are not in the programme
If your organisation isn't part of Project Glasswing, there is nothing here you can call, and nothing you are missing: Fable 5.1 is the generally available model with the same capability. The separate id exists for access control, not because the model underneath behaves differently.
Related
See claude-mythos-5-1 for the successor, and claude-fable-5 for the full specification this model shares in every respect except its access path.
Verified 2026-09-30 against claude-api skill — shared/models.md, model-migration.md + SKILL.md model table and Anthropic's pricing and model-deprecation pages (canonical: https://platform.claude.com/docs/en/about-claude/models/overview, https://platform.claude.com/docs/en/about-claude/pricing, https://platform.claude.com/docs/en/about-claude/model-deprecations).
Verified 2026-10-28 against Anthropic's model-deprecations status table and pricing page (canonical: https://platform.claude.com/docs/en/about-claude/model-deprecations, https://platform.claude.com/docs/en/about-claude/pricing).Could not confirm: These dates apply to the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Microsoft Foundry. Amazon Bedrock and Claude Models on Google Cloud (Google managed) keep their own retirement schedules, which we do not track.