Rankings / Anthropic
claude-opus-4-5-20251101 via api.anthropic.com
legacy. Superseded through Opus 4.6/4.7/4.8 and then Opus 5; still served under the 60-day-notice deprecation policy. Successor: Claude Opus 5 (claude-opus-5, released 2026-07-24).
Usage share 0.03% · OpenRouter rankings API (daily token share, 2026-08-04)
What happens to your data: training-on-customer-data defaults, retention windows, residency options, and the enterprise-versus-consumer terms gap. A legal property, not a capability — it survives every model generation.
API and enterprise inputs are not used for training by default, with documented retention controls. Note the consumer gap: claude.ai consumer accounts default to training opt-in (with opt-out) since late 2025.
Whether it changes without warning: versioning discipline, changelog quality, deprecation policy, and observed silent changes. The signal no one else tracks.
Dated model snapshots and a documented deprecation policy; release notes exist but behavior-affecting changes to system prompts and safety filters are not always changelogged.
“This page lists all API deprecations, along with recommended replacements.”
Whether an attacker can make it misbehave — direct jailbreaks against the model's own policies and indirect prompt injection in agentic tool use. Graded to the weaker of the two, because an attacker takes the easier path.
Constitutional-classifier safeguards and extensive internal plus external red-teaming; consistently among the top scorers in independent jailbreak-resistance testing.
Strong relative performance in Gray Swan agent red-teaming (run with UK AISI), but indirect injection in multi-step tool use still succeeds at non-trivial rates — as it does for every frontier model.
Whether you can see how it was built and tested: model cards, published safety evals, external pre-deployment testing, and disclosure of changes. The mechanism behind the tier ladder.
Detailed system cards, the published Responsible Scaling Policy with ASL levels, external pre-deployment testing (UK AISI, third-party evaluators), and a transparency hub.
“Based on our assessments, we have decided to deploy Claude Opus 4.5 under the ASL-3 Standard.”
Whether it is certified and compliant: SOC 2, ISO/IEC 42001, HIPAA eligibility, EU AI Act readiness, and audit availability.
SOC 2 Type 2, ISO 27001, ISO/IEC 42001 certification, HIPAA-eligible configurations; audit documentation via trust portal.
Regional pinning available — customers can pin processing to a chosen region.
US default on the first-party API; regional options via Bedrock and Vertex.
claude.ai consumer accounts default to training opt-in (with opt-out) since late 2025, while API and enterprise inputs carry a no-train default.
3 tracked changes. The typical (median) interval between them is ~150 days (low confidence: a median of only 2 intervals). The last change was on 2026-07-24, 12 days before the as-of date (2026-08-05) — about 0.1× the typical interval. That is within its historical cadence.
No dated score readings recorded for this model yet. Readings are only entered where multiple real, dated third-party values exist — never interpolated.
retrieved 2026-08-03 — 2026-08-05
System card names UK AISI, Trajectory Labs, 10a Labs, and Gray Swan; includes an adverse capability finding (agentic cyber-range success against weakly-secured networks) published against interest. Day-one availability on Bedrock, Vertex, and Foundry. The Opus 4.5 entry moves to legacy.
Release accompanied by a detailed system card including third-party evaluation results — the disclosure pattern Tier 2 requires.
Consumer claude.ai accounts were transitioned to a training-permitted default with a five-year retention window (opt-out available). API and enterprise tiers unchanged. Widens the enterprise/consumer gap tracked under the data-handling vector.
Compare this model: Claude Opus 4.5 + open compare view →