At Box, we evaluated Claude Opus 5 on the Box Complex Work Eval — our benchmark of realistic, document-grounded tasks across twelve industries, benchmarked here against the prior-generation Claude Opus 4.8. The tasks mirror the analytical workthat knowledge workers actually do: reading source documents, reconciling numbers, running due diligence, and reviewing expert output for errors. Here’s what we found.
Opus 5 scores 68% overall to Opus 4.8’s 63% — a 5-point step up on complex enterprise work, and it leads on a clear majority of tasks. The gains are widest exactly where the work is hardest: exhaustive, multi-step analysis over dense source material.
Consistent across every type of analytical work
Opus 5 leads across the modes of analytical work, and its edge is largest on the most demanding one: due diligence (the exhaustive, find-everything work) where it scores 76% to Opus 4.8’s 65%, an 11-point gap. It also improves on report drafting (69% vs 67%), data analysis (60% vs 54%), and expert review (67% vs 66%). The through-line: Opus 5 stays accurate as the number of required findings grows, rather than catching the obvious items and stopping.
Stronger where the work is most technical
The gains concentrate in the most technically demanding industries:
- Legal (79% vs 71%): On a clause-by-clause contract review against a policy, Opus 5 scored every item completely and correctly cleared the clauses that were acceptable under an exception, where Opus 4.8 missed items
- Technology (67% vs 56%): On a multi-input computation across markets, it applied the logic consistently and produced values that tied out
- Healthcare (65% vs 58%): On an expert-document review, it caught more errors and classified each by severity correctly, rather than under-detecting
- Life sciences (63% vs 49%): Its signature win was on intersecting several datasets under a strict matching rule to find the items common to all, where Opus 4.8 over-included partial matches and dropped genuine ones
The more a task depends on exhaustive, precise reasoning over multiple documents, the larger Opus 5’s advantage.
Get started
On complex enterprise knowledge work, Claude Opus 5 is a clear step up from the prior generation — its advantage concentrated on the exhaustive, multi-step analysis that drives real decisions. Opus 5 is available now in Box AI. Reach out to your Box account team or try it in Box AI Studio.
