Multimodal Code Generation on 3DCodeBench
78.3ExecutabilityPairCoder
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PairCoderBase Model=gpt-5.4-mini, Reasoning Status=thinking off, Collaboration Framework=PairCoder2026.07 | 78.3 | 70.2 | 37.6 | 2.62 | |
| gpt-5.4-miniBase Model=gpt-5.4-mini, Reasoning Status=thinking off, Collaboration Framework=single2026.07 | 20 | 18.1 | 10.1 | 4.85 |