InternVL3_5-4B vs LFM2.5-VL-3B

OpenGVLab

Wins 0 of 2 areas

—

Liquid AI · released Aug 11, 2026

Wins 2 of 2 areas

Images and charts · Following instructions

LFM2.5-VL-3B is the stronger all-rounder.

Scores updated Sep 19, 2026 · 27 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.

Images and chartsUnderstanding pictures, charts and video13LFM2.5-VL-3B3 of 4 tests
Following instructionsDoing exactly what it is asked01LFM2.5-VL-3B1 of 1 test

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 LFM2.5-VL-3B pulls ahead

Harder college exam questions with imagesMMMU-Pro+7.8points aheadLFM2.5-VL-3B30.5InternVL3_5-4B22.7
Visual perception tasks people solve at a glanceBLINK+4.3points aheadLFM2.5-VL-3B61.5InternVL3_5-4B57.2
Math problems shown in pictures and chartsMathVista+1.4points aheadLFM2.5-VL-3B68.5InternVL3_5-4B67.1

Where InternVL3_5-4B pulls ahead

Questions that truly need the image to answerMMStar+2.2points aheadInternVL3_5-4B65.5LFM2.5-VL-3B63.3

Every test, side by side

All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.

Images and chartsLFM2.5-VL-3B

MMMU-ProLFM2.5-VL-3B by 7.822.730.5+7.8
BLINKLFM2.5-VL-3B by 4.357.261.5+4.3
MMStarInternVL3_5-4B by 2.265.563.3+2.2
MathVistaLFM2.5-VL-3B by 1.467.168.5+1.4

Full images and charts ranking

Following instructionsLFM2.5-VL-3B

IFBenchLFM2.5-VL-3B by 1.324.525.8+1.3

Full following instructions ranking

Other results22 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.

IFEvalLFM2.5-VL-3B by 46.935.482.3+46.9
MMMU (val) (Pass@1)InternVL3_5-4B by 12.360.748.4+12.3
MMEInternVL3_5-4B by 7.98173.1+7.9
TextVQA-valLFM2.5-VL-3B by 6.877.584.3+6.8
InfographicVQA (val)InternVL3_5-4B by 6.776.970.2+6.7
ScreenSpot-v2 MobileInternVL3_5-4B by 6.687.881.2+6.6
MM-IFEvalLFM2.5-VL-3B by 6.154.560.6+6.1
RealWorldQALFM2.5-VL-3B by 5.467.773.1+5.4
ChartQA TestInternVL3_5-4B by 4.986.281.3+4.9
HallusionBenchInternVL3_5-4B by 4.952.147.2+4.9
ScreenSpot-v2 DesktopInternVL3_5-4B by 3.38278.7+3.3
Multilingual MMBenchLFM2.5-VL-3B by 2.976.679.5+2.9
OCRBench v1LFM2.5-VL-3B by 2.28284.2+2.2
SimpleVQALFM2.5-VL-3B by 1.733.735.4+1.7
OCRBench v2_enInternVL3_5-4B by 1.649.147.5+1.6
MMMBLFM2.5-VL-3B by 1.581.583+1.5
Multilingual MMMBLFM2.5-VL-3B by 1.581.583+1.5
SEED-Bench (image)LFM2.5-VL-3B by 1.376.477.7+1.3
DocVQA-valtie91.891.1tie
ScreenSpot-v2 Webtie82.682.2tie
POPEtie88.988.7tie
MMBenchtie81.181tie

Questions people ask

Which is better, InternVL3_5-4B or LFM2.5-VL-3B?

LFM2.5-VL-3B wins all two areas where both have results: images and charts and following instructions. InternVL3_5-4B wins none.

How do you compare the two?

We use the 27 benchmark tests both models have published scores on. The verdict counts the 5 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 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.