Claude 3 Haiku Release Notes

Anthropic has released Claude 3 Haiku, the fastest and most affordable model in its intelligence class. This model is designed for enterprise applications that require rapid analysis of large datasets and low-latency output for tasks such as customer support.

High-Speed Token Processing

Claude 3 Haiku is three times faster than its peers for the majority of workloads. For prompts under 32K tokens, the model can process 21K tokens (approximately 30 pages of text) per second. This speed enables the execution of numerous small tasks in tandem and creates more responsive, engaging chat experiences.

Enterprise Cost Efficiency

Claude 3 Haiku utilizes a 1:5 input-to-output token pricing ratio specifically designed for enterprise workloads involving long prompts. The model is available at half the cost of other models in its performance tier, allowing businesses to analyze large volumes of documents—such as legal cases, contracts, or quarterly filings—more economically. For one US dollar, Claude 3 Haiku can process and analyze 400 Supreme Court cases or 2,500 images.

Vision Capabilities and Use Cases

Claude 3 Haiku includes state-of-the-art vision capabilities, making it a versatile solution for processing visual data alongside text. Its combination of speed and affordability makes it suitable for high-volume enterprise tasks, including the analysis of thousands of images or extensive legal documentation.

Security and Robustness

Anthropic employs rigorous testing to minimize the risk of jailbreaks and harmful outputs. The model's security framework includes:

• Continuous systems monitoring
• Endpoint hardening
• Secure coding practices
• Strong data encryption protocols
• Stringent access controls
• Regular security audits and collaboration with penetration testers

Availability

Claude 3 Haiku is available via the Anthropic API and for Claude Pro subscribers on claude.ai. It is also available on Amazon Bedrock and is scheduled for release on Google Cloud Vertex AI.