Claude Haiku 4.5 API pricing & cost

Current token rates, context limits, workload examples and unit-economics shortcuts for Claude Haiku 4.5.

Calculate my workload →Official source ↗

Input / 1M tokens$1

Output / 1M tokens$5

Cached input / 1M$0.1

Context window200K tokens

Pricing page fallback verified September 3, 2026; client checks TokenCOGS live registry on load.

Quick cost examples

What the token rates mean in dollars

Example workload: 2,000 input tokens + 500 output tokens per request, standard text-token pricing.

Volume	Estimated API cost	Approx. / request
1,000 requests	$4.50	$0.0045
100,000 requests	$450.00	$0.0045
1,000,000 requests	$4,500.00	$0.0045

Change the token assumptions →

Positioning

Where Claude Haiku 4.5 sits on price

Claude Haiku 4.5 sits toward the lowest input-token rate in the current TokenCOGS registry for this provider. Price alone is not a quality ranking, but this position is useful when you are screening models for unit-economics fit before benchmarking latency and quality.

Cost comparison is not a quality, latency or reliability ranking. Benchmark the task you actually need to run.

Compare all Anthropic models →

Cost considerations

What can move the invoice

Cached input is tracked at $0.1 per 1M tokens, which can materially change repeated-system-prompt or repeated-context workloads.

Current deals & credits

Ways this provider may reduce spend

Credits program

API credits + priority limits

Anthropic Startup Program

See details ↗

Compare

Nearby Anthropic options

Model	Input / 1M	Output / 1M	Context
Claude Sonnet 5	$2	$10	1M tokens	View
Claude Sonnet 4.6	$3	$15	Not listed in registry	Calculate
Claude Sonnet 4.5	$3	$15	Not listed in registry	Calculate
Claude Opus 5	$5	$25	1M tokens	View
Claude Opus 4.8	$5	$25	Not listed in registry	Calculate

Use your workload

Turn pricing into unit economics

Cost per userEstimate AI COGS per active customer →Agent costModel multi-step tasks and retries →RAG costAdd retrieval context and vector spend →

FAQ

Claude Haiku 4.5 pricing questions

How much does Claude Haiku 4.5 cost per 1M tokens?

The current TokenCOGS snapshot lists Claude Haiku 4.5 at $1 per 1M input tokens and $5 per 1M output tokens. Live registry data can update these values when the provider changes pricing.

What would 1,000 Claude Haiku 4.5 requests cost?

At 2,000 input tokens and 500 output tokens per request, 1,000 requests are approximately $4.50 before provider-specific extras. Change the workload in the linked calculator for your own usage pattern.

Is Claude Haiku 4.5 the cheapest Anthropic model?