Anthropic on Tuesday launched Claude Opus 5.5, a new AI model that delivers performance comparable to its top-tier Fable 5.1. However, the company says it costs 40 percent less to run than its predecessor. This makes the new model a more affordable option for developers who need high-level performance.
The release comes amid growing debate over AI risks. Earlier this month, CEO Dario Amodei called on the global AI community to slow the pace of releasing new capabilities. Therefore, safety has become a central theme around this launch.
Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family.
— Claude (@claudeai) September 22, 2026
It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5. pic.twitter.com/Q9C2VKQ79f
Anthropic said Opus 5.5 underwent external testing by independent AI safety research groups Frontier Design and METR before its release. Additionally, the model includes safeguards that were previously reserved for Anthropic’s most capable systems. According to the company, Opus 5.5 also outscored OpenAI’s GPT-5.6 Sol on a software development benchmark. Meanwhile, it cost roughly one-third as much to run for that same task.
Pricing for the model stands at $4 per million input tokens and $20 per million output tokens. This marks a 20 percent drop compared to Opus 5. Furthermore, the model is available on major platforms, including Amazon Web Services, Google Cloud and Microsoft Azure.
On safety, the model performed better than earlier systems in internal testing. It was about 85 percent less likely than Opus 5 or Mythos 5.1 to attempt bypassing containment boundaries during a dedicated evaluation. Instead, the company says the added safeguards helped reduce this kind of behaviour significantly.
Anthropic also confirmed that Claude Sonnet 5.5 and Claude Haiku 5.5 will launch in the coming weeks. These releases are expected to bring many of the same performance, speed and safety improvements. Also, the company is expanding programs that give vetted cybersecurity and life-sciences researchers broader access to its models.
Still, questions remain about how the industry will balance rapid AI development with safety concerns. Finally, Opus 5.5 offers an early signal of how Anthropic plans to address that balance going forward.
