PCFBench - Product Decomposition: leaderboard
Metric: Bill-of-materials F1 (%; decomposing each of 94 EPD products into constituent materials; from PCFBench, a product-carbon-footprint benchmark of 614 expert-labelled items; scored deterministically or by a temperature-0 judge; higher is better). Source: arxiv.org. Saturation forecast: Around June 2027. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 80 |
| 2 | Claude Opus 4.6 | 78 |
| 3 | Gemini 3.1 Pro (Preview) | 74 |
| 4 | Claude Sonnet 4.6 | 74 |
| 5 | DeepSeek V3.2 | 74 |
| 6 | GPT-5.5 | 73 |
| 7 | Claude Haiku 4.5 | 73 |
| 8 | GPT-5.4 Mini | 70 |
Interactive version: theaggregate.ai/benchmark?slug=pcfbench-product-decomposition · How It Works · Data refreshed daily, snapshot 2026-09-26.