Archive · 23 Sep 2026
Anthropic Launches Claude Opus 5.5 Amid AI Slowdown Debate
Anthropic released a new artificial-intelligence model called Claude Opus 5.5.
It is the first model in Anthropic’s new 5.5 family.
The company says it is about as capable as its top-tier Fable 5.1 model on many tasks.
Anthropic also says Opus 5.5 performs better than Fable 5.1 on several tests.
It costs less to use than the older Opus 5 model and produces answers more than 30% faster.
The model has safeguards for sensitive cybersecurity and biology-related capabilities.
In one company test, it was about 85% less likely to try to get around containment boundaries.
Outside groups including Frontier Design and METR tested it before release.
The launch comes as Anthropic Chief Executive Dario Amodei has called for slower development of new AI capabilities because of safety concerns.
Anthropic launched Claude Opus 5.5 as the first model in its new 5.5 family.
The company says it matches Claude Fable 5.1 on most tasks and exceeds it on several benchmarks.
Anthropic says typical workload costs are about 40% lower than Opus 5, with pricing of $4 per million input tokens and $20 per million output tokens.
Output generation is reported to be more than 30% faster than Opus 5, with improvements in communication and response organization.
The model includes stronger safeguards and was about 85% less likely than Opus 5 or Claude Mythos 5.1 to attempt to bypass containment boundaries in one evaluation.
- Who
- Anthropic released Claude Opus 5.5; Chief Executive Dario Amodei has called for caution about releasing new AI capabilities.
- What
- A new AI model was launched with claimed improvements in performance, speed, cost efficiency and safety.
- When
- Anthropic launched the model on Tuesday; it is available immediately, while Sonnet 5.5 and Haiku 5.5 are expected in the coming weeks.
- Where
- It is available through the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure.
- Why
- Anthropic presented the model as a more capable and efficient system while applying stronger safeguards amid debate over AI risks.
Capability Expansion
Safety and Caution
Pace of AI development
Capability Expansion
Anthropic is releasing a more capable model and says it can deliver stronger performance, faster output and lower operating costs.
Safety and Caution
The launch comes amid safety concerns, and Chief Executive Dario Amodei has called on the AI community to slow the release of new capabilities.
Access to sensitive capabilities
Capability Expansion
Anthropic is expanding access for vetted cybersecurity and life-sciences researchers, while offering a verification program for legitimate research.
Safety and Caution
The model launches with safeguards that restrict assistance with tasks such as discovering exploits in compiled software or developing biological weapons.
Evidence of safety
Capability Expansion
Anthropic says Opus 5.5 performed strongly in its internal behavioural audit and was less likely to attempt to bypass containment boundaries.
Safety and Caution
The safety claims are based primarily on Anthropic’s evaluations, although external groups including Frontier Design and METR also tested the model before launch.
- Model
- Claude Opus 5.5, the first model in Anthropic’s 5.5 family.
- Reported performance
- Comparable to Claude Fable 5.1 on most tasks and stronger on several listed benchmarks, according to Anthropic.
- Standard pricing
- $4 per million input tokens and $20 per million output tokens.
- Cost reduction
- Typical workload costs are approximately 40% lower than with Opus 5.
- Speed improvement
- Output generation is reported to be more than 30% faster than Opus 5.
- Safety evaluation
- About 85% less likely than Opus 5 or Claude Mythos 5.1 to attempt to bypass containment boundaries in one evaluation, according to Anthropic.
- Availability
- Available through the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure.
