Anthropic · Very Fast · Updated 2026-11

Claude 3.5 Haiku API Pricing & Cost Breakdown

Low-latency interactive UX, code autocomplete, content moderation, and sub-agent tasks.

Input / 1M Tokens
$0.80
Cached Input / 1M
$0.080
Output / 1M Tokens
$4.00
Context Window
200K

Workload Cost Scenarios (30-Day Billing)

Supports 90% prompt caching discount ($0.08/1M cached read).

Scenario In / Out per Req Uncached Monthly 50% Cache Hit Monthly
Hobby / Prototype (100 req/day) 1500 / 400 $8.40 $6.78
Growing Startup (2,500 req/day) 2500 / 700 $360.00 $292.50
Production Scale (50,000 req/day) 3000 / 800 $8,400.00 $6,780.00

Compare Claude 3.5 Haiku Against Alternatives

Lower-Cost Alternative Models

Gemini 2.0 Flash ($0.40/1M out) Qwen 2.5 72B ($0.40/1M out) Grok 3 Mini ($0.50/1M out) GPT-4o mini ($0.60/1M out)

Frequently Asked Questions

What is Claude 3.5 Haiku API pricing per 1M tokens?

Standard input is $0.80/1M, cached input is $0.080/1M, and output is $4.00/1M.

What is the context window of Claude 3.5 Haiku?

Claude 3.5 Haiku supports a 200K token context window with up to 8,192 max output tokens.

How much does prompt caching save on Claude 3.5 Haiku?

Prompt caching reduces cached input token rates by 90% compared with uncached input.

What workloads is Claude 3.5 Haiku best suited for?

Low-latency interactive UX, code autocomplete, content moderation, and sub-agent tasks.