Anthropic launches Claude Opus 5.5

Anthropic launched Claude Opus 5.5 and described it as the first model in a new family. It said Claude Sonnet 5.5 and Claude Haiku 5.5 would follow in the coming weeks.

The company said Opus 5.5 matched Claude Fable 5.1 on most tasks while costing about 40 per cent less to run than Opus 5. Anthropic set the price at $4 per million input tokens and $20 per million output tokens, down from $5 and $25.

Anthropic said cache read costs were down 60 per cent and that total operating costs should be about 40 per cent lower than before. It also said output was more than 30 per cent faster and that subscriber usage limits would rise.

The company said the model wrote more naturally, used less jargon and followed instructions more closely. It also said the model added safeguards for cybersecurity, biology and frontier LLM development, and that some flagged requests would be routed to other models.

Anthropic said verified organisations could apply for biological research access through the Life Sciences Verification Program, and that the Cyber Verification Program would be extended to Opus 5.5 in the coming weeks. Frontier Design and METR tested the model before release, and Anthropic said it was available on Amazon Web Services, Google Cloud and Microsoft Azure. Artificial Analysis said Opus 5.5 scored 58 on its Intelligence Index, led six of ten evaluations and tied GPT-6 Astra on Terminal-Bench 4.0.