{T}
TokenMath.net
Official Provider HubSan Francisco, CA, USA

Anthropic API Pricing & Token Economics

Anthropic develops the Claude model family (Opus, Sonnet, Haiku, and Fable). Claude models feature industry-leading coding benchmarks and prompt caching with a 90% read discount. Caching is explicitly controlled using ephemeral cache breakpoints with a 5-minute time-to-live (TTL) that auto-renews on every cache hit.

Prompt Caching Policy

90% read discount with 1.25x write surcharge on 1,024+ token prefixes (5-minute auto-refreshing TTL).

Batch / Asynchronous Rules

50% off standard input and output rates via the Anthropic Message Batches API.

Special Billing Conditions

Zero write fees on standard prefix evaluation; automatic cache invalidation.

Active Anthropic Models(4 models verified)

All 30 Models →
ModelContextInput / 1MCached / 1MOutput / 1MBatch DiscountActions
Claude Fable 5.1
claude-fable-5-1-20260820
1M$10.00$0.25$50.0050% offSpecs →
Claude Opus 5
claude-opus-5-20260815
1M$5.00$0.50$25.0050% offSpecs →
Claude Sonnet 5
claude-sonnet-5-20260901
1M$2.00$0.20$10.0050% offSpecs →
Claude Haiku 4.5
claude-haiku-4-5-20260515
200k$1.00$0.10$5.0050% offSpecs →

Calculate your monthly Anthropic bill

Simulate prompt caching hit rates, batch discounts, and output token costs with zero server logging.

Open Cost Calculator →