Anthropic Claude Platform API pricing (serverless, per MTok, 2026-10-04 list price)

USD per MILLION TOKENS (MTok), LIST PRICE, observed 2026-10-04. Current lineup, standard: Claude Fable 5.1 input $10 / output $50 / cache write $12.50 / cache read $0.25; Opus 5.5 $4 / $20 / $5 / $0.20; Sonnet 5.5 $2 / $10 / $2.50 / $0.20; Haiku 4.5 $1 / $5 / $1.25 / $0.10. price_usd = Opus 5.5 input. Batch tier is exactly half of standard on every row (Fable 5.1 $5/$25, Opus 5.5 $2/$10, Sonnet 5.5 $1/$5, Haiku 4.5 $0.50/$2.50). Fast mode is 2x standard (Opus 5.5 $8/$40). Legacy rows retained on the page for comparison: Opus 5 / 4.8 / 4.7 / 4.6 / 4.5 all $5/$25 input-output; Sonnet 4.6 and 4.5 and 4 all $3/$15; Haiku 3.5 $0.80/$4; Opus 4.1 and 4 $15/$75 PERISHABLE-FIGURE STAMP (added 2026-10-04, uniform across data/supply/ so it is greppable). PRICE CLASS: PUBLISHED LIST PRICE - an asking rate a new customer is quoted, NOT a realised, discounted or negotiated transaction rate. A real bill is at or below this and typically well below it once commitments, credits and committed-use discounts apply. OBSERVED: 2026-10-04 RE-VERIFY BY: 2027-01-04 - the repo's quarterly re-verification bar. After that date this figure is presumed drifted and must be re-checked against its source before it is quoted again. A price without its observation date is noise; this stamp exists so a reader can tell a fresh figure from a stale one without reading prose. UNIT: USD per MILLION TOKENS (MTok), LIST PRICE, observed 2026-10-04 READING RULE: No further perishable flag is published for this figure.

lead_time_weeks is null because NO NUMBER APPLIES, and the reason is specific rather than generic: this record prices a metered service term (tokens, egress, storage) rather than hardware capacity, so there is no order-to-delivery interval to report at all. Calling this 'no published lead time' would be wrong in the other direction - nothing was withheld.

no EAR citation: no controlled item is transferred by this transaction; not an EAR citation: provider service terms and end-use policy (the AI Diffusion Rule briefly tiered service access before its 2025-05-13 rescission)

MECHANISM: a token-priced API, so no accelerator crosses a border and the 3A090 hardware licence regime does not attach to the transaction. Export control becomes an END-USER AND TERMS question instead: the provider applies geographic and end-use restrictions to service access, and a customer can still be refused under sanctions or end-use policy wherever the compute sits. A reader targeting a restricted destination should read the provider's terms, not the EAR.

Anthropic's cache-read discount is the most aggressive in the industry and it is NOT a flat 10%: the page states a cache hit costs 10% of standard input price generally, 2.5% on Fable 5.1 / Mythos 5.1, and 5% on Opus 5.5 - so a 2.5% cache read on Fable makes caching pay off after a single read, while on Sonnet 5.5 the 10% rule needs one read to break even on the 5-minute TTL. Three cost levers a reader will otherwise miss: (a) cache pricing reflects the 5-minute TTL, and the 1-hour TTL has its own (higher write) price - model your cache lifetime before assuming the headline rate; (b) US-only inference is priced at 1.1x input and output for workloads that must run in the US; (c) Anthropic bills through AWS Marketplace and Azure Marketplace as Claude Consumption Units at a FIXED $0.01 per CCU, with any negotiated discount applied at the token-to-CCU conversion and NOT to the CCU price - so a marketplace buyer sees one line item and cannot see their effective per-token rate in the invoice. Session-runtime metering also exists on the Platform ($0.08 per session-hour plus tokens), which is a different billing shape from raw API calls. No GPU hourly rate and no region list are published. EXPORT CONTROL (2026-10-04): NOT APPLICABLE IN THE HARDWARE-SHIPMENT SENSE - this record prices a token-based API, so no accelerator is transferred across a border and the ECCN 3A090 hardware licence regime does not attach to the transaction. Recorded null rather than false precisely because a token API is where export control instead becomes an END-USER/TERMS question: the provider applies geographic and end-user restrictions to service access (the AI Diffusion Rule briefly put a tiered country structure on exactly this kind of service before being rescinded on 2025-05-13), and a customer can still be refused under sanctions or end-use policy regardless of where the compute sits. Anyone reading this record for a restricted destination should look at the provider's terms, not at the EAR. Sources for the regime: spl-bis-2025-05-ai-diffusion-rescission, spl-bis-2026-05-d5-guidance. EXPORT CONTROL (2026-10-04): NOT APPLICABLE IN THE HARDWARE-SHIPMENT SENSE - this record prices a token-based API, so no accelerator crosses a border and the ECCN 3A090 hardware licence regime does not attach to the transaction. Recorded NULL rather than false precisely because a token API is where export control instead becomes an END-USER and TERMS question: the provider applies geographic and end-user restrictions to service access (the AI Diffusion Rule briefly put a tiered country structure on exactly this kind of service before being rescinded and unenforced from 2025-05-13), and a customer can still be refused under sanctions or end-use policy regardless of where the compute sits. For a restricted destination, read the provider's terms, not the EAR. Sources: spl-bis-2025-05-ai-diffusion-rescission, spl-bis-2026-05-d5-guidance. LEAD TIME: null - correct and not a gap. A token-priced serverless API is provisioned on request, so there is no lead time to publish; the binding constraint when a model is at capacity is provider-side queuing and any per-account concurrency limit, which this repo records in notes where a vendor publishes one (see the DeepSeek concurrency figures already here). Deliberately NOT recorded: any 'model capacity lead time' estimate, because vendors publish none and the SEO 'GPU lead time' numbers that circulate do not apply to a rented endpoint. Probe log in source spl-negative-no-oem-publishes-lead-time.