Anthropic
Claude Fable / Opus, explicit cache breakpoints, 90% read discount.
How caching works here
You mark up to four cache breakpoints with cache_control. Everything before a breakpoint is cached. Reads are 90% off, but the first write costs 25% more than list input, so a breakpoint only pays off once the block is reused enough times to amortise the write.
What Zumik sees
Anthropic gives the highest read discount in the catalog, and our traces show the best median capture (around 90%+) - but only when breakpoints sit on genuinely stable blocks. Zumik places them from snapshot structure rather than guesswork.
Putting a breakpoint after volatile content, or breaking on every turn, means you pay write premiums repeatedly and never recover them.
Anthropic models in the catalog.
| Model | Context | Input | Output | Cache read | Reuse-adj |
|---|---|---|---|---|---|
| Claude Haiku 4.5 | 200K | $1.00 | $5.00 | $0.10 −90% | $1.63 |
| Claude Sonnet 4.5 | 200K | $3.00 | $15.00 | $0.30 −90% | $4.89 |
| Claude Sonnet 4.6 | 1M | $3.00 | $15.00 | $0.30 −90% | $4.89 |
| Claude Opus 4.5 | 200K | $5.00 | $25.00 | $0.50 −90% | $8.14 |
| Claude Opus 4.6 | 1M | $5.00 | $25.00 | $0.50 −90% | $8.14 |
| Claude Opus 4.7 | 1M | $5.00 | $25.00 | $0.50 −90% | $8.14 |
| Claude Opus 4.8 | 1M | $5.00 | $25.00 | $0.50 −90% | $8.14 |
| Claude Fable 5 | 1M | $10.00 | $50.00 | $1.00 −90% | $16.29 |
| Claude Opus 4.1 | 200K | $15.00 | $75.00 | $1.50 −90% | $24.43 |
Anthropic, answered.
How does Anthropic prompt caching work?
You mark up to four cache breakpoints with cache_control. Everything before a breakpoint is cached. Reads are 90% off, but the first write costs 25% more than list input, so a breakpoint only pays off once the block is reused enough times to amortise the write.
What discount does Anthropic caching give?
Cache reads on Anthropic are about 90% cheaper than list input price, though the first write costs about 25% more than list input.
Does Anthropic support BYOK on Zumik?
Yes. You can bring your own Anthropic key, and provider-native caching, batch, and service tiers stay active under your account.
What is the common Anthropic caching mistake?
Putting a breakpoint after volatile content, or breaking on every turn, means you pay write premiums repeatedly and never recover them.
Route Anthropic the smart way.
Capture Anthropic's 90% cache-read discount and batch tier automatically through Zumik.