All dimensions
G 88CJ 84Sc 83Num 80Risk 85Q 84EU 80Use 86
European B2B software investment benchmark
Investment judgement,
benchmarked.
Not “can the model answer?” but “can the model help decide?”
Methodology v0.3
Synthetic standings as an evaluation ledger.
12 models / 8 dimensions / closed-book
All dimensions
G 86CJ 82Sc 84Num 82Risk 82Q 82EU 78Use 84
All dimensions
G 84CJ 78Sc 76Num 83Risk 78Q 80EU 74Use 80
All dimensions
G 85CJ 76Sc 74Num 81Risk 77Q 79EU 72Use 82
All dimensions
G 83CJ 75Sc 73Num 80Risk 75Q 77EU 70Use 80
All dimensions
G 81CJ 73Sc 71Num 79Risk 73Q 76EU 70Use 78
All dimensions
G 80CJ 72Sc 70Num 78Risk 71Q 75EU 69Use 77
All dimensions
G 79CJ 71Sc 70Num 77Risk 72Q 74EU 66Use 75
All dimensions
G 78CJ 70Sc 68Num 76Risk 70Q 72EU 64Use 74
All dimensions
G 76CJ 67Sc 66Num 74Risk 68Q 70EU 60Use 72
All dimensions
G 75CJ 65Sc 64Num 73Risk 66Q 69EU 58Use 71
All dimensions
G 74CJ 63Sc 62Num 75Risk 64Q 67EU 56Use 69
| Delta | Model | Mode | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | new | OAIGPT-5.5OpenAI | 83.9 | 88.0 | 84.0 | 83.0 | 80.0 | 85.0 | 84.0 | 80.0 | 86.0 | Closed-book |
| 2 | ▼ 1 | Claude Opus 4.8Anthropic | 82.8 | 86.0 | 82.0 | 84.0 | 82.0 | 82.0 | 82.0 | 78.0 | 84.0 | Closed-book |
| 3 | ▲ 3 | Gemini 3.5 FlashGoogle | 79.5 | 84.0 | 78.0 | 76.0 | 83.0 | 78.0 | 80.0 | 74.0 | 80.0 | Closed-book |
| 4 | ▼ 1 | OAIGPT-5.4OpenAI | 78.4 | 85.0 | 76.0 | 74.0 | 81.0 | 77.0 | 79.0 | 72.0 | 82.0 | Closed-book |
| 5 | • | Claude Opus 4.7Anthropic | 76.9 | 83.0 | 75.0 | 73.0 | 80.0 | 75.0 | 77.0 | 70.0 | 80.0 | Closed-book |
| 6 | ▼ 2 | Gemini 3.1 ProGoogle | 75.2 | 81.0 | 73.0 | 71.0 | 79.0 | 73.0 | 76.0 | 70.0 | 78.0 | Closed-book |
| 7 | ▲ 1 | Grok 4.20xAI | 74.1 | 80.0 | 72.0 | 70.0 | 78.0 | 71.0 | 75.0 | 69.0 | 77.0 | Closed-book |
| 8 | ▲ 2 | ZGLM-5.2Z.ai | 73.4 | 79.0 | 71.0 | 70.0 | 77.0 | 72.0 | 74.0 | 66.0 | 75.0 | Closed-book |
| 9 | • | Qwen3.7 MaxAlibaba | 71.9 | 78.0 | 70.0 | 68.0 | 76.0 | 70.0 | 72.0 | 64.0 | 74.0 | Closed-book |
| 10 | ▲ 1 | Seed 2.0 ProByteDance | 69.6 | 76.0 | 67.0 | 66.0 | 74.0 | 68.0 | 70.0 | 60.0 | 72.0 | Closed-book |
| 11 | ▼ 3 | DeepSeek V4 ProDeepSeek | 68.1 | 75.0 | 65.0 | 64.0 | 73.0 | 66.0 | 69.0 | 58.0 | 71.0 | Closed-book |
| 12 | new | OAIGPT-5.2 ProOpenAI | 66.8 | 74.0 | 63.0 | 62.0 | 75.0 | 64.0 | 67.0 | 56.0 | 69.0 | Closed-book |
Scoring shape
Judgement carries the most weight.
A grounded but commercially naive model should not lead an investment benchmark. Overall is derived from eight dimension scores, never stored directly.
Grounding
15%Claims are supported by packet evidence or clearly marked as inference.
Commercial judgement
20%Identifies what matters economically and strategically to the deal.
Scepticism
15%Challenges weak claims, inflated metrics and convenient narratives.
Numerical sanity
15%Interprets and reconciles financial and operating data correctly.
Risk discovery
15%Finds hidden and second-order risks across the packet.
Question generation
10%Asks the next questions most likely to change the recommendation.
European context
5%Accounts for fragmented markets, regulation, language and procurement.
Output usefulness
5%Produces material a deal team can reuse with limited editing.
Interpretation
Workflow usefulness, not general intelligence.
The current board is a product demonstration built from synthetic scores. The methodology, case structure and data contracts are the testable artefacts.
