MBA-Bench - Technical Validity: leaderboard
The technical validity dimension of MBA-Bench (2026): the technical feasibility of a generated business idea, rated by the MLLM judge from 1 to 4; higher is better.
Metric: Technical Validity rating (1-4; technical feasibility of the idea, scored by an InternVL2.5-78B MLLM judge over five business ideas for each of three questions per test image on the MBA-Bench multimodal business-ideation test set). Source: arxiv.org. Saturation forecast: Not forecast. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 3.16 | #158 |
| 2 | GPT-5 | 3.08 | #163 |
| 3 | GPT-5 Mini | 3.07 | #349 |
| 4 | Qwen 2.5 VL 7B Instruct | 3.06 | #1842 |
| 5 | Gemini 3.5 Flash | 3.05 | #98 |
| 6 | GPT-4o | 3.04 | #741 |
| 7 | Gemini 3.1 Pro (Preview) | 3.04 | #91 |
| 8 | Gemini 3.6 Flash | 3.03 | #72 |
| 9 | Qwen 2.5 VL 32B Instruct | 2.99 | #1055 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mba-bench-technical-validity · How It Works · Data refreshed daily, snapshot 2026-11-12.