Claude Fable5 pricing, in real numbers
Claude Fable5 costs $10 in, $50 out per million tokens — double Claude Opus 4.8 and Claude Opus 5, which both list at $5/$25. Use the calculator to see what that means for your workload.
Input
$10
per million tokens
Output
$50
per million tokens
Prompt caching cuts cached input tokens by 90% — long system prompts and specs get dramatically cheaper on repeat calls.
- Context window
- 1M tokens
- Max output
- 128K tokens
- Thinking
- Adaptive only — temperature / top_p removed
- Tier
- Mythos-class — priced a tier above Opus
Monthly cost calculator
Estimate a month of usage and compare across the current Claude lineup.
| Model | API price (in / out) | Estimated monthly cost | vs Claude Fable5 |
|---|---|---|---|
| Claude Fable5 | $10 / $50 | $1,000 | — |
| Claude Opus 4.8 | $5 / $25 | $500 | $-500 |
| Claude Sonnet 5 | $2 / $10 | $200 | $-800 |
| Claude Sonnet 4.6 | $3 / $15 | $300 | $-700 |
| Claude Haiku 4.5 | $1 / $5 | $100 | $-900 |
Claude Sonnet 5 uses its introductory rate of $2 / $10 per million tokens (through August 31, 2026; standard $3 / $15 after). All current models — Sonnet 5, Fable 5 and Opus 5 — share the tokenizer introduced with Opus 4.7, so the figures below compare on equal terms. Only estimates carried over from Opus 4.6 or earlier need a ~30% upward adjustment.
The full API price table
The $10/$50 headline is only the start — caching, batch, and the safeguard fallback all change what you actually pay.
| Token type | Claude Fable5 | Claude Opus 4.8 |
|---|---|---|
| Base input | $10 | $5 |
| Output | $50 | $25 |
| 5-min cache write | $12.50 | $6.25 |
| 1-hour cache write | $20 | $10 |
| Cache hits & refreshes | $1 | $0.50 |
| Batch input (−50%) | $5 | $2.50 |
| Batch output (−50%) | $25 | $12.50 |
Caching and Batch API discounts stack: cached batch input lands around $0.50 per million tokens — roughly 95% off list price.
Safeguard fallback bills at Opus rates: when a cybersecurity or biology classifier reroutes a request to Opus 4.8, that response is charged at $5/$25 — never at Fable5 prices.
US-only inference costs 1.1x the standard rate on both input and output, for workloads that must stay in the United States.
Subscription access timeline
- July 20, 2026 onward — Max and Team Premium permanently include Fable5 for up to 50% of weekly usage limits. Pro and Team Standard typically use usage credits.
- Usage credits bill at API rates ($10/$50 per M tokens) when inclusion does not apply or the 50% Fable share is exhausted. API/Bedrock/OpenRouter invoices are always separate.
Pricing FAQ
Quick answers on what Claude Fable5 actually costs.
Build on Claude Fable5 with your eyes open
Know the Claude Fable5 cost math, the safeguard behavior, and the right tier before you commit. Create a free account to get notified when specs, prices, or access channels change.