SupChain-Bench (Tool Calling, With SOP): leaderboard
Metric: Information retrieval accuracy (%) on SupChain-Bench's 98 order-diagnosis questions answered by calling tools over a simulated supply-chain order database, a question counting as correct when the order information its tool calls retrieve matches what an oracle run of the expert procedure gathers; the prompt adds the expert standard operating procedure, step by step; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Mini | 86.73 | #176 |
| 2 | GPT-5 | 84.69 | #91 |
| 3 | GPT-5 Nano | 84.69 | #415 |
| 4 | Claude 3.7 Sonnet | 40.09 | #241 |
| 5 | Claude Sonnet 4 | 39.79 | #194 |
| 6 | Gemini 2.5 Pro | 37.56 | #145 |
| 7 | Claude Opus 4 | 34.69 | #155 |
| 8 | Claude 3.5 Sonnet | 29.59 | #337 |
| 9 | GPT-4.1 Mini | 28.57 | #346 |
| 10 | Qwen 3 Max | 28.57 | #201 |
| 11 | Gemini 2.5 Flash | 23.46 | #237 |
| 12 | DeepSeek R1 | 20.2 | #245 |
| 13 | GPT-4o | 16.32 | #333 |
| 14 | GPT-4.1 | 16.32 | #240 |
| 15 | QwQ-32B | 16.32 | #410 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=supchain-bench-tool-calling-with-sop · How It Works · Data refreshed daily, snapshot 2026-10-11.