MidMarketBench
Rank 5Synthetic score

Claude Opus 4.7

Provider
Anthropic
Version
4.7
Release
2026-02
Context
Provider listed
Rank
#5
Overall
76.8
Run mode
Closed-book
Weights
Closed

Roster source: provider documentation

Dimension shape

Where the score comes from.

The profile is finance-weighted. Commercial judgement breaks ties before grounding.

Grounding83
Commercial judgement75
Scepticism73
Numerical sanity80
Risk discovery75
Question generation77
European context70
Output usefulness80

Board distribution

Position in the current synthetic field.

6586

Representative outputs

Useful judgement versus polished summary.

These are benchmark examples, not claims about unrun model behaviour. Live attribution begins only when run artefacts exist.

strong sample

GPT-5.5

G 90CJ 88Sc 89Num 86Risk 90Q 86EU 84Use 88

Ranked diligence risks

  1. NRR quality: Headline NRR of 118 percent is not broad-based. FY25 expansion of €1.9m includes roughly €1.1m from one automotive account and €0.5m from CPI uplift, leaving about €0.3m of seat and module expansion. Action: decompose expansion by account, module, seat and price over three years.
  2. Services contamination: Services are about 28 percent of revenue at roughly 20 percent gross margin. Action: restate software-only economics and test whether implementation can be productised.
  3. Customer concentration: Account A is 11.7 percent of ARR; the top ten are 41.8 percent. Action: review renewal dates, contract protection and switching costs for the top ten.
  4. Geographic repeatability: Product-market fit is DACH-proven. UK and Benelux remain early and fragmented. Action: separate pipeline, win rate, implementation cost and retention by country.
  5. TAM overreach: The €14bn figure is a broad category. The reachable market is regulated mid-market manufacturers with multi-site audit workflows, plausibly €0.6bn to €0.9bn. Action: rebuild from target sites and ACV.

The central question is whether durable, usage-led expansion exists once price and one-off enterprise effects are removed.

Evaluator rationale: Grounds each issue in the packet, ranks by decision impact and proposes a specific diligence action.

weak sample

GPT-5.2 Pro

G 62CJ 55Sc 52Num 58Risk 57Q 60EU 54Use 64

Key risks

  1. Competition in the compliance software market is intense.
  2. Customer churn is a risk for SaaS businesses.
  3. There is key-person risk in the founder-led structure.
  4. Macroeconomic conditions in Europe could affect growth.
  5. Execution risk exists around the AI roadmap.

Overall, the company shows strong momentum with 118 percent NRR and a large €14bn market opportunity.

Evaluator rationale: Lists generic risks, accepts headline NRR and TAM, and misses the packet's most decision-relevant contradictions.