Model · claude-haiku-4-5

Claude Haiku 4.5

Use Claude Haiku 4.5 when a workflow needs low-cost Anthropic capacity for high-volume background units.

standard

confidential

responses

200k ctx

1 provider

Lane pricing

Provider rates, live status, and capability fit for this model.

input $0.4795, cached input $0.0479, output $2.4 per 1M tokens from currently available batch discounted customer rates.

Catalog model

No advanced runtime capability has been declared.

No provider-hosted web search, code execution, retrieval, or other tools declared.

Why use this model

Use it for responses jobs where cost, routing auditability, and async delivery matter more than single-request latency.

The catalog declares a 200k context window, which helps reject lanes that cannot fit large batch items before dispatch.

Available privacy tiers include standard, confidential, so routing can enforce buyer constraints.

Per-lane input, cached-input, and output rates are shown in the page instead of buried in a raw JSON payload.

Providers

Best p50

—

Context

200k

Cost estimate in <200ms. Final charge follows delivered token usage.