Claude Opus 5.5
Anthropic's September 2026 frontier model for coding, computer use and knowledge work. The hosted release reaches Anthropic's platform and major clouds, while classifier and fallback safeguards restrict high-risk cyber and biology requests. Its underlying model is tested in a detailed system card; model-artifact licensing was not independently established in this run.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why this model scores 87.7
Capability 96 and autonomy 96 reflect the exact version's strong coding, tool-use and long-horizon evaluations, including 66.4% on Terminal-Bench 4.0, with vendor-benchmark limits. Deployment 97 reflects immediate hosted access across Anthropic and major clouds, not open weights. Misuse 85 reflects high underlying cyber and biology capability but substantial deployed classifier, fallback and verification restrictions. Control difficulty 62 reflects stronger controlled containment and prompt-injection results, balanced against evaluation awareness, residual uncertainty and the challenge of monitoring long autonomous tasks. Availability remains unknown until affirmative artifact-licensing evidence is reviewed; hosted access alone is not proof of proprietary artifact status.
News tied to Claude Opus 5.5
The model score of 87.7 rates this model's risk profile. The overall Doom Index of 67.9 measures the complete temporally weighted evidence record. These values answer different questions.
Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.
Anthropic releases Claude Opus 5.5 with stronger agentic performance and high-risk safeguards
Anthropic released Claude Opus 5.5 for hosted use across its platform and major clouds. Its published coding and agentic evaluations show gains over Opus 5, while the system card reports strong cyber and biology capabilities. Anthropic applies classifier-based restrictions and model fallback in high-risk domains. Its improved containment-boundary results were obtained in controlled evaluations, not a real-world escape test.
- Full item contribution
- +0.12
- Claude Opus 5.5 equal share
- +0.12
Model score history
-
R1
Doom Score 87.7
New exact version, dated official release and linked 230-page Anthropic system card, not present in fresh durable model catalogue.
23 Sept 2026