Anthropic releases Claude Opus 5.5 at a lower price

AI-generated illustrationAI-generated illustration
Anthropic released Claude Opus 5.5 on 22 September 2026, available immediately through its own platform and through AWS, Google Cloud and Microsoft Azure. Pricing is $4 per million input tokens and $20 per million output tokens, against $5 and $25 for Opus 5, and cache reads fall from $0.50 to $0.20 per million. Anthropic puts the saving at about 40% on typical workloads, with output generated more than 30% faster.
On Anthropic’s figures, Opus 5.5 scores 66.4% on the Terminal-Bench 4.0 agentic coding benchmark, against 52.3% for Opus 5, and 1846 Elo on the GDPval-AA v2.1 knowledge-work evaluation, against 1708. The company says it is comparable to Claude Fable 5.1, its larger model, on most work, and notes that benchmark margins have become a less reliable guide to real-world differences.
Cybersecurity and biology safeguards are similar to those applied to Fable 5.1, with unrestricted access in those areas gated behind verification programmes. The release is the second frontier model of the day: OpenAI launched GPT-6 Sol and Luna on the same date, also on lower prices.