Anthropic Launches Claude Opus 5.5 at 60% Lower Cost
In brief
- Claude Opus 5.5 matches Fable 5.1 performance at 60% lower cost
- Input tokens cost $4/M (down from $5), output $20/M (down from $25)
- Opus 5.5 leads Terminal-Bench 4.0 and FrontierCode; GPT-6 Astra wins two benchmarks
- Model includes Fable 5.1-level safeguards due to Mythos-class capabilities
- Available now on AWS, Google Cloud, and Microsoft Azure
Pricing and Performance Claims
Input tokens now run $4 per million for Opus 5.5 versus $5 for Opus 5, while output tokens drop to $20 from $25. Cache reads—reusing context a model has already processed instead of reprocessing it from scratch—cost $0.20 per million tokens, a 60% reduction. The pricing shift reflects Anthropic's effort to push faster, cheaper inference without sacrificing capability.
Anthropic says the new model performs at the level of Claude Fable 5.1 on most tasks. On some published benchmarks, Opus 5.5 does beat it outright. On Terminal-Bench 4.0, Opus 5.5 hit 66.4%, ahead of Fable 5.1's 55.8%. On FrontierCode, Opus 5.5 scored 54.4% versus Fable 5.1's 50.3%. On GDPval-AA v2.1, Opus 5.5 scored 1846 against Fable 5.1's 1735.
Where GPT-6 Astra Leads
That said, OpenAI's GPT-6 Astra still edges Opus 5.5 on two key tests. GPT-6 Astra's 41.4% edges out Opus 5.5's 40.0% on AutomationBench. On Terminal-Bench-Science 0.1, Astra's 64.6% beats Opus 5.5's 58.7% by a wider margin. The gap isn't dramatic, but it's there—and it matters for teams benchmarking against the latest.
Because Opus 5.5 matches Anthropic's Mythos-class models (the company's most restricted tier, reserved for vetted cybersecurity and defense use), it launches with the same safeguards as Fable 5.1. Opus 5.5 is live now across Anthropic's platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure.
The Bigger Picture
CEO Dario Amodei published an essay calling on the industry to slow down, arguing that AI capability gains are outrunning safety progress. The timing—releasing a faster, cheaper model while urging restraint—signals Anthropic's balancing act: ship competitive products while publicly advocating for measured development.
For teams already on Opus 5 or Fable 5.1, the math is straightforward: Opus 5.5 costs less and performs better on most tasks. Whether it's enough to sway teams locked into OpenAI's ecosystem remains an open question.
