InternVL3_5-2B wins more areas, narrowly.
Scores updated · 29 tests both models report · How we compare
Where each one wins
Tests won in each of the two areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
Following instructions rests on a single test.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where InternVL3_5-2B pulls ahead
Every test, side by side
All 29 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Images and chartsEven
Following instructionsInternVL3_5-2B
- IFBenchInternVL3_5-2B by 3.624.420.8+3.6
Other results23 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- ScreenSpot-v2 MobileInternVL3_5-2B by 78.686.27.6+78.6
- ScreenSpot-v2 WebInternVL3_5-2B by 77.479.92.5+77.4
- ScreenSpot-v2 DesktopInternVL3_5-2B by 73.979.96+73.9
- IFEvalLFM2-VL-3B by 40.532.472.9+40.5
- RealWorldQALFM2-VL-3B by 9.561.671.1+9.5
- MMMU (val) (Pass@1)InternVL3_5-2B by 6.45245.6+6.4
- TextVQA-valLFM2-VL-3B by 6.476.683+6.4
- MMMBLFM2-VL-3B by 5.676.381.9+5.6
- Multilingual MMMBLFM2-VL-3B by 5.676.381.9+5.6
- Multilingual MMBenchLFM2-VL-3B by 5.470.976.3+5.4
- MM-IFEvalLFM2-VL-3B by 4.347.151.4+4.3
- MMBenchLFM2-VL-3B by 3.876.280+3.8
- SimpleVQALFM2-VL-3B by 2.530.533+2.5
- OCRBench v1InternVL3_5-2B by 2.283.981.7+2.2
- MMBench (dev en)LFM2-VL-3B by 1.678.279.8+1.6
- OCRBench v2_enInternVL3_5-2B by 1.645.543.9+1.6
- InfographicVQA (val)InternVL3_5-2B by 1.569.367.8+1.5
- DocVQA-valLFM2-VL-3B by 1.488.489.8+1.4
- ChartQA TestInternVL3_5-2B by 1.381.780.4+1.3
- HallusionBenchInternVL3_5-2B by 1.247.646.4+1.2
- POPELFM2-VL-3B by 1.28889.2+1.2
- SEED-Bench (image)LFM2-VL-3B by 1.275.476.6+1.2
- MMEtie73.673tie
Questions people ask
Which is better, InternVL3_5-2B or LFM2-VL-3B?
InternVL3_5-2B wins one of the two areas where both have results: following instructions. LFM2-VL-3B wins none. They are level on images and charts.
How do you compare the two?
We use the 29 benchmark tests both models have published scores on. The verdict counts the 6 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 23 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.