Anthropic has released Claude Opus 5.5, a new AI model designed to provide higher performance and lower operating costs while addressing user feedback regarding writing style and efficiency.
Key facts
- •Opus 5.5 costs 40 percent less to run than its predecessor, Opus 5.
- •The model generates output more than 30 percent faster than previous versions.
- •Artificial Analysis gave the model a record-high score of 58 on its Intelligence Index.
- •The model includes new anti-distillation measures and EU AI Act-compliant watermarking.
- •Anthropic plans to release Sonnet 5.5 and Haiku 5.5 in the coming weeks.
Anthropic has launched Claude Opus 5.5, the first in a new family of AI models. The company reports that the model matches the performance of Claude Fable 5.1 on most tasks while running faster and costing significantly less than its predecessor, Opus 5. The release aims to address customer feedback regarding cost, efficiency, and communication quality, particularly for users in legal, financial, and software development sectors.
By the numbers
Pricing and Performance Improvements
Anthropic has reduced token prices for Opus 5.5 to $4 per million input tokens and $20 per million output tokens, representing a 20 percent decrease from Opus 5. Additionally, cache read costs have been reduced by 60 percent. The company states that total operating costs are approximately 40 percent lower, and the model generates output more than 30 percent faster. Subscribers will also see a 20 percent increase in five-hour usage limits.
Communication and Safety Features
The new model is designed to communicate more naturally by prioritizing key information and reducing jargon, addressing previous criticism of 'Claudish' writing. Opus 5.5 also includes cybersecurity, biology, and frontier LLM development safeguards that match those of Fable 5.1. To combat 'distillation attacks,' the model includes 'Preserved Thinking' measures and watermarking to comply with the EU AI Act.
Benchmarking and Availability
Independent platform Artificial Analysis reports that Opus 5.5 achieved a score of 58 on its Intelligence Index, the highest recorded. The model is currently available on platforms including Amazon Web Services, Google Cloud, and Microsoft Azure. Anthropic plans to release Sonnet 5.5 and Haiku 5.5 in the coming weeks.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by The Decoder.
