← Back to Test Cases
According to 21 CFR 211.186, what specific information must be included in master production and control records?
GMP KnowledgeGeneral Gmphardretired
This question has been retired from scoring. Every model answered it correctly on every attempt, so it no longer separates them. It stays published for reference, and historical results against it remain in the database.
Cross-Model Comparison
| Model | Score | Latency | Tokens In | Tokens Out |
|---|---|---|---|---|
| GPT-5.4 | 100.0% | 2.6s | 102 | 134 |
| GPT-5.4 mini | 100.0% | 959ms | 102 | 52 |
| GPT-5.4 nano | 100.0% | 962ms | 102 | 68 |
| Claude Haiku 4.5 | 100.0% | 2.9s | 108 | 269 |
| Claude Sonnet 4.6 | 100.0% | 8.5s | 108 | 357 |
| Llama 4 Scout | 100.0% | 740ms | 106 | 235 |
| Claude Opus 4.6 | 100.0% | 12.4s | 108 | 391 |
| Llama 4 Maverick | 100.0% | 17.4s | 106 | 371 |
| Llama 3.3 70B Instruct | 100.0% | 2.1s | 107 | 194 |
| Mistral Small 2603 | 100.0% | 1.9s | 118 | 301 |
| Qwen3.5-397B-A17B | 100.0% | 27.7s | 112 | 1,796 |
| DeepSeek-V3.2 | 100.0% | 17.2s | 101 | 182 |
| DeepSeek-R1 | 100.0% | 10.1s | 107 | 323 |
| DeepSeek-R1-Distill-Qwen-32B | 100.0% | 100.4s | 104 | 646 |
| Mistral Large 3 675B | 100.0% | 4.4s | 106 | 269 |
| DeepSeek V4 Flash | 100.0% | 4.0s | 101 | 421 |
| Qwen3.5-35B-A3B | 100.0% | 6.8s | 112 | 1,199 |
| Gemini 3.1 Pro | 100.0% | 19.5s | 105 | 1,034 |
| Gemini 3 Flash | 100.0% | 3.5s | 104 | 389 |
| Gemini 3.1 Flash-Lite | 100.0% | 2.5s | 104 | 294 |
| Qwen3.6 27B | 100.0% | 22.9s | 112 | 1,627 |
| Qwen3.6 35B A3B | 100.0% | 8.1s | 112 | 1,416 |
| DeepSeek V4 Pro | 100.0% | 14.0s | 101 | 775 |
| DeepSeek-V3.2 | 100.0% | 8.0s | 101 | 151 |
| Gemma 4 26B A4B IT | 100.0% | 8.5s | 117 | 356 |
| Gemma 4 31B IT | 100.0% | 16.3s | 117 | 359 |
| MiniMax M2.7 | 100.0% | 3.1s | 134 | 293 |
| Claude Opus 4.8 | 100.0% | 8.1s | 146 | 577 |
| GLM-5.2 | 100.0% | 13.0s | 108 | 684 |
Tags
batch_recordsmaster_recordsdocumentation