Claude Opus 5

A dated reading of what is claimed, reported, and independently verified in the current evidence.

Current reading

Readiness band and full integer triple

AI-assisted assembly · derived results

Readiness band

Mature

Primary summary from verified readiness

Evidence strength Critical

Decision

Human editorial judgment · 2026-07-30

Adopt with guardrails

Why: Anthropic's own pricing table confirms the vendor's 'half the price of Fable 5' claim exactly ($5/$25 vs $10/$50 per MTok), and both official and press sources corroborate near-Fable-5 coding/agent benchmark parity (CursorBench 3.2 within 0.5%, OSWorld 2.0 ahead at 1/3 cost) — but the same sources disclose concrete capability gaps in cybersecurity work that must be respected before broad or security-adjacent adoption.
Next: Pilot claude-opus-5 as the model backing conductor- and implementer-tier agents (per feedback_ultracode_agent_models: currently pinned to plain 'opus') for a cost/perf trial given its documented half-of-Fable-5 pricing, then gate any security-sensitive or exploit-adjacent automation behind Mythos-5/human review since Opus 5 explicitly trails on cybersecurity exploitation and cannot scan binaries.

Constraints

No named blocker is present in the current public projection.

Evidence summary

AI-assisted assembly

Counts and dates only. Raw signals, private excerpts, trust records, and internal corpus material are not published here.

Publication record

Initial public reading