Other · model
LFM2.5-VL-3B benchmark scores
As of 2026-11-22, LFM2.5-VL-3B (Other) has measured scores on 10 benchmarks and estimated scores on 24 more.
| Benchmark | Score | Source |
|---|---|---|
| GDPval-AA | 0.0% | estimated ± 5.5 pp, low confidence |
| IFBench | 25.8% | measured |
| IFEval | 82.3% | measured |
| τ²-bench results | 55.4% | estimated ± 12.4 pp, low confidence |
| AA-MMMU-Pro | 30.5% | estimated ± 1.5 pp, medium confidence |
| BFCL v4 | 32.5% | measured |
| AA-IFBench | 56.7% | estimated ± 7.4 pp, low confidence |
| MMMU | 48.4% | measured |
| ParseBench-1 | 32.4% | estimated ± 8.4 pp, low confidence |
| ParseBench-2 | 19.4% | estimated ± 5.0 pp, low confidence |
| CharXiv | 61.9% | estimated ± 3.3 pp, medium confidence |
| CharXiv w/o tools | 69.8% | estimated ± 1.2 pp, medium confidence |
| AI2D_TEST | 87.6% | estimated ± 1.4 pp, medium confidence |
| CountBench | 87.3% | measured |
| DynaMath | 79.2% | estimated ± 1.5 pp, low confidence |
| ERQA | 39.9% | estimated ± 4.2 pp, medium confidence |
| MathVision | 34.6% | estimated ± 3.0 pp, medium confidence |
| MLVU (M-Avg) | 64.5% | estimated ± 2.2 pp, medium confidence |
| MMMU-Pro | 30.5% | measured |
| OmniDocBench 1.5 | 85.5% | estimated ± 1.8 pp, low confidence |
| RealWorldQA | 73.1% | measured |
| ScreenSpot Pro | 37.0% | estimated ± 4.8 pp, low confidence |
| SimpleVQA | 35.4% | measured |
| V* | 50.2% | estimated ± 5.8 pp, low confidence |
| Video-MME (w/o subtitle) | 60.2% | estimated ± 1.2 pp, medium confidence |
| Video-MME (with subtitle) | 53.5% | estimated ± 3.0 pp, medium confidence |
| VideoMMMU | 43.6% | estimated ± 2.3 pp, medium confidence |
| We-Math | 61.3% | estimated ± 5.5 pp, low confidence |
| MedXpertQA (MM) | 65.0% | estimated ± 7.1 pp, low confidence |
| MMVU | 39.5% | estimated ± 2.7 pp, low confidence |
| CC-OCR | 80.1% | estimated ± 0.7 pp, medium confidence |
| RefCOCO (avg) | 87.9% | measured |
| OCRBench V2 | 47.5% | measured |
| PinchBench | 34.6% | estimated ± 5.0 pp, low confidence |