Activefastllm Proprietary

Claude Haiku 4.0

by Anthropic· Released May 2025· Cutoff April 2025

Claude Haiku 4.0 is Anthropic's fastest and most cost-effective model in the Claude 4 family, designed for high-throughput, low-latency applications. It offers a strong balance of speed and intelligence, making it ideal for simple queries, classification, and real-time chat. As the successor to Claude 3 Haiku, it delivers improved performance across benchmarks while maintaining competitive pricing.

Official Site API Docs

Input cost

$0.80 per 1M tokens

Output cost

$4.00 per 1M tokens

Context window

200K tokens

Max output

8192 tokens

Modalities

textimage

License

proprietary

Capabilities

Function CallingVisionCode GenerationStreamingJSON ModeTool UseMultilingual Support

Best For

High-speed, cost-sensitive applications requiring low latency and reliable performance.

Strengths

Fastest inference speed in the Claude 4 family
Lowest cost per token among Claude 4 models
Strong performance on classification and routing tasks
Supports vision and tool use despite small size

Limitations

Lower reasoning capability compared to Claude 4 Sonnet and Opus
Not suitable for complex multi-step tasks
Smaller context window than Claude 4 Sonnet (200K vs 200K, same)
May produce less nuanced responses on creative tasks

Use Cases

Real-time customer support chatbots

Content classification and moderation

Simple data extraction and summarization

High-volume API call routing

Lightweight code generation and debugging

Multilingual translation tasks

Image captioning and basic visual Q&A

Improvements Over Previous Model

Context window increased from 100K to 200K tokens
Max output tokens doubled from 4096 to 8192
Significantly faster inference — ~2× lower latency than Claude 3 Haiku
Improved coding benchmark scores (HumanEval 88% vs 75%)
Added native vision/image input support
Lower pricing: $0.80/M tokens vs $1.00/M tokens for input
Enhanced multilingual capabilities with better non-English support

Back to all models

Improvements Over Previous Model

Context window increased from 100K to 200K tokens

Max output tokens doubled from 4096 to 8192

Significantly faster inference — ~2× lower latency than Claude 3 Haiku

Improved coding benchmark scores (HumanEval 88% vs 75%)

Added native vision/image input support

Lower pricing: $0.80/M tokens vs $1.00/M tokens for input

Enhanced multilingual capabilities with better non-English support