Foundation Model · Amazon BedrockHaiku FamilyLightweight TierText

Claude 3.5 Haiku

Amazon Bedrock token pricing and cost optimization reference for Claude 3.5 Haiku.Rates normalized to USD per 1 Million Tokens ($/1M) with prompt caching and batch API discounts across 3 AWS regions.

ProviderAnthropic

Maker HQSan Francisco, California, USA

Class / TierLightweight

ModalityText

Prompt CachingSave 90%

Batch APISave 50%

AWS Regions3 Global

Bedrock Model IDanthropic.claude-3-5-haiku-v1:0

Input / Prompt

$0.80/ 1M tokens

$0.0008 per 1K tokens

Output / Completion

$4.00/ 1M tokens

$0.0040 per 1K tokens

Prompt Cache Read

$0.08/ 1M tokens

Save 90% vs prompt

Batch Inference

$0.40/ 1M in

Save 50% via Batch API

Price Ratio

1 : 5.0

Prompt to completion factor

Regional Reach

3AWS Regions

Compare 3 regions ↓

On Google Cloud Vertex AI

Vertex AI isn't shown for Claude 3.5 Haiku: Anthropic's models are priced there per geography and context band, and this catalog publishes Claude on Amazon Bedrock only.

Alternatives to Claude 3.5 Haiku

Option	Price vs Claude 3.5 Haiku	Why look at it
Claude Instant Compare →	inputsame price output-40%-$1.60/1M Standard tier, per 1M tokens	Next step down in price from Anthropic (general tier)
Claude Haiku 4.5 Compare →	input+25%+$0.20/1M output+25%+$1.00/1M Standard tier, per 1M tokens	Next step up in price from Anthropic
Claude 3 Haiku Compare →	input-69%-$0.55/1M output-69%-$2.75/1M Standard tier, per 1M tokens	Same Haiku line, different version
Claude Sonnet 5.5 Compare →	input+150%+$1.20/1M output+150%+$6.00/1M Standard tier, per 1M tokens	Also from Anthropic (flagship tier)

Claude 3.5 Haiku Spend Estimator

Model inference spend by prompt volume, cache hit rate, and batch processing.

Prompt Tokens (Millions / month)

M tokens

Completion Tokens (Millions / month)

M tokens

Prompt Cache Hit Rate (90% discount on cached context)%

Batch API Mode50% discount on non-realtime async jobs via S3

Estimated Monthly Spend

$0.00

Fresh Prompts:$0.00

Cached Prompts:$0.00

Completions:$0.00

Active: Input: $0.80/1M · Output: $4.00/1M

Key Cost Levers & Architecture

Practical mechanisms to reduce inference costs with Claude 3.5 Haiku on AWS

Prompt CachingSave ~90%

Caches repetitive system instructions, schemas, and document prefixes in memory with 5-minute and extended 1-hour TTL tiers.

Read Rate: $0.08 / 1M tokens
Write Rate (5m TTL): $1.00 / 1M tokens
Ideal for: Multi-turn conversations, RAG document queries, structured schemas

Batch InferenceSave 50%

Asynchronous bulk processing submitted via Amazon S3 with a standard 24-hour turnaround SLA.

Batch Input: $0.40 / 1M tokens
Batch Output: $2.00 / 1M tokens
Ideal for: Offline classification, synthetic data generation, benchmark evals

Cross-Region & Tiers3 Regions

AWS Bedrock supports standard on-demand inference, priority reserved throughput, and cross-region routing.

Standard: Lowest latency, pay-per-token with zero commitments
Cross-Region Routing: Automatically burst traffic to available regional capacity
GovCloud: Dedicated isolated compliance partitions with +20% uplift

Rates by region

Live rates across all 3 available AWS cloud regions

Baseline: $0.80/1M in3 Global Regions

AWS Region	Prompt Input $/1M	Completion Output $/1M	Cache Read $/1M	Batch Input $/1M	Pricing Tier
us-east-1 US East (N. Virginia)	$0.80	$4.00	$0.08	$0.40	Baseline
us-west-2 US West (Oregon)	$0.80	$4.00	$0.08	$0.40	Baseline
us-east-2 US East (Ohio)	$0.80	$4.00	$0.08	—	Baseline

Developer Quickstart

Invoke Claude 3.5 Haiku via Amazon Bedrock Converse API

Python (boto3) · Amazon Bedrock Runtime

import boto3

# Amazon Bedrock Converse API invocation
client = boto3.client("bedrock-runtime", region_name="us-east-1")

response = client.converse(
    modelId="anthropic.claude-3-5-haiku-v1:0",
    messages=[
        {
            "role": "user",
            "content": [{"text": "Summarize key features and cost levers of Claude 3.5 Haiku."}]
        }
    ],
    inferenceConfig={
        "maxTokens": 1024,
        "temperature": 0.7
    }
)

output_text = response["output"]["message"]["content"][0]["text"]
usage = response["usage"]
print(f"Response: {output_text}")
print(f"Usage: {usage['inputTokens']} in, {usage['outputTokens']} out")