Anthropic's Claude Haiku 5.5 Launches on AWS with Lower Prices and Higher Performance

Category: AI Models

Anthropic announced the release of Claude Haiku 5.5, now available on Amazon Web Services (AWS). The new model demonstrates significant improvements, scoring 72.4 percent on the OSWorld computer use benchmark while offering token price reductions of up to 90 percent.

Performance and Benchmarks

Claude Haiku 5.5 shows notable progress in its ability to act as a computer agent, achieving a score of 72.4 percent on the OSWorld benchmark. This test measures the model's proficiency in performing tasks within a simulated computer operating system. Additionally, analysis suggests that Claude Haiku 5.5 performs better than competitors like GPT-6 Luna at the same pricing, offering a strong combination of capability and cost-effectiveness.

Pricing and Availability

A key aspect of this release is a substantial price reduction, with token prices dropping by up to 90 percent. This makes the model more accessible for a wider range of applications, from chatbots to content summarization. The model is immediately available for use on the AWS platform, allowing developers to integrate it into their existing cloud infrastructure. More details can be found in the official AWS announcement.

Expanded Access for Enterprise Use

In a related development, Anthropic has also provided security teams with access to Claude models under fewer safety restrictions. This change is intended to support specialized applications in areas like cybersecurity, where models may need to analyze potentially malicious code or content for defensive purposes.

Additional sources: - latent.space - the-decoder.com

Source article: the-decoder.com