B2V-Bench (With Few-Shot Examples): leaderboard
B2V-Bench (Peking University and Alibaba, 2026) consumer-value identification from Taobao purchase behaviour, with a prompt that adds the behavioural value descriptions and labelled few-shot examples. Macro-F1 in percent over 28 value labels plus none on 544 expert-labelled episodes; higher is better.
Metric: Macro-F1 (%; multi-label classification of the consumer values, 28 labels of the E-commerce Consumption Value Taxonomy plus none, expressed in each of 544 Taobao purchase-decision episodes labelled from scratch by psychology experts; the model reads the GPT-5.4-written summary of the pre-purchase behaviour log; prompt adds the behavioural value descriptions and few-shot examples). Source: arxiv.org. Saturation forecast: Around December 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 41.7 | #163 |
| 2 | GLM-5 | 35.8 | #256 |
| 3 | DeepSeek V3.2 | 35 | #403 |
| 4 | Kimi K2.6 | 30.4 | #173 |
| 5 | Qwen 3 235B A22B | 28.8 | #622 |
| 6 | Claude Haiku 4.5 | 27.5 | #568 |
| 7 | GPT-4o Mini | 22 | #1627 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=b2v-bench-with-few-shot-examples · How It Works · Data refreshed daily, snapshot 2026-11-12.