Multilingual Reading Comprehension on Belebele
79.8AccuracySP3F-7B
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SP3F-7BLanguage=vi2026.01 | 79.8 | 100 | — | — | — | — | |
| SP3F-7BLanguage=nl2026.01 | 77.9 | 100 | — | — | — | — | |
| SP3F-7BLanguage=af2026.01 | 72.5 | 100 | — | — | — | — | |
| SP3F-7BLanguage=tr2026.01 | 69.1 | 100 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=nl2026.01 | 67.2 | 99.1 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=vi2026.01 | 66.5 | 99.8 | — | — | — | — | |
| SP3F-7BLanguage=he2026.01 | 64.1 | 99.2 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=he2026.01 | 61.3 | 72.5 | — | — | — | — | |
| SP3F-7BLanguage=tl2026.01 | 60.5 | 100 | — | — | — | — | |
| SP3F-7BLanguage=Average2026.01 | 58.3 | 99.9 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=tr2026.01 | 49.7 | 98.3 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=Average2026.01 | 39.9 | 93.4 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=gu2026.01 | 30.6 | 99.5 | — | — | — | — | |
| SP3F-7BLanguage=pa2026.01 | 23 | 100 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=af2026.01 | 19.2 | 98.8 | — | — | — | — | |
| SP3F-7BLanguage=gu2026.01 | 19.2 | 100 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=tl2026.01 | 16.6 | 81.3 | — | — | — | — | |
| Qwen2.5-7B-InstructLanguage=pa2026.01 | 8.4 | 97.8 | — | — | — | — | |
| GLM-4.5VZero-shot=true, Pass@1=true, Parameters=106B-A12B, Organization=Z.ai2026.05 | — | — | 94.3 | 87.26 | 74.08 | 90.95 | |
| Llama 4 MaverickZero-shot=true, Pass@1=true, Parameters=400B-A17B, Organization=MetaAI2026.05 | — | — | 92.85 | 90.84 | 83.91 | 92.07 | |
| Nemotron 3 SuperZero-shot=true, Pass@1=true, Parameters=120B-A12B, Organization=Nvidia2026.05 | — | — | 95.2 | 88.04 | 69.94 | 91.62 | |
| Phoenix-VL 1.5 MediumZero-shot=true, Pass@1=true, Parameters=123B, Organization=HTX2026.05 | — | — | 95.64 | 92.74 | 84.47 | 93.3 | |
| Qwen 3.5Zero-shot=true, Pass@1=true, Parameters=122B-A10B, Organization=Alibaba Cloud2026.05 | — | — | 95.75 | 91.51 | 82.79 | 93.52 |