Claude Opus 4.1

What the provider states

Autonomy & replication · ASL-4 · Stated as not crossed
since claude opus 4.1 is deployed with asl-3 protections, we focused on confirming it remains well below asl-4 thresholds across cbrn (chemical, biological, radiological, and nuclear), cyber, and autonomy domains.
Anthropic · August 2025 · Read the source
Chemical, biological, radiological and nuclear · ASL-4 · Stated as not crossed
since claude opus 4.1 is deployed with asl-3 protections, we focused on confirming it remains well below asl-4 thresholds across cbrn (chemical, biological, radiological, and nuclear), cyber, and autonomy domains.
Anthropic · August 2025 · Read the source
Cyber · ASL-4 · Stated as not crossed
since claude opus 4.1 is deployed with asl-3 protections, we focused on confirming it remains well below asl-4 thresholds across cbrn (chemical, biological, radiological, and nuclear), cyber, and autonomy domains.
Anthropic · August 2025 · Read the source

Safety training (harmlessness and refusal)
to address misuse concerns, we implemented the same safeguards, including pre-deployment measures such as harmlessness training and updating the computer use instructions to emphasize appropriate usage.
Anthropic · August 2025 · Read the source

long-form virology task 1 · The provider · Chemical, biological, radiological and nuclear
long-form virology task 1: claude opus 4.1 scored 0.86 (0.98 for sequence design, 0.87 for protocol design), closely matching claude opus 4’s score of 0.84 (0.98 for sequence design, 0.85 for protocol design).
Anthropic · August 2025 · Read the source
long-form virology task 2 · The provider · Chemical, biological, radiological and nuclear
long-form virology task 2: claude opus 4.1 scored 0.67 (0.85 for sequence design, 0.79 for protocol design), demonstrating slightly poorer performance on our viral sequence design task compared to claude opus 4’s 0.72 (0.91 for sequence design, 0.79 for protocol design).
Anthropic · August 2025 · Read the source
malicious applications of computer use · The provider · Injection & agentic
as before, we evaluated the model’s willingness and ability to comply with malicious requests that violate our usage policy when given access to computer use capabilities.
Anthropic · August 2025 · Read the source
single-turn benign request evaluations · The provider · Harmful content
on benign requests touching potentially sensitive topics, claude opus 4.1 showed performance comparable to that of claude opus 4.
Anthropic · August 2025 · Read the source

No document we read states: a release date, an independent or government test. No incident on record names this version.

Some documents we hold are stored in lower case only, so their quotes are shown in lower case. The source link opens the original.

Documents listed for this version