LFM2-VL-3B vs Qwen3.5 4B

Liquid AI · released Oct 22, 2025

Wins 0 of 2 areas

—

Alibaba · released Mar 2, 2026

Wins 2 of 2 areas

Images and charts · Following instructions

Qwen3.5 4B is the stronger all-rounder.

Scores updated Sep 19, 2026 · 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.

Images and chartsUnderstanding pictures, charts and video05Qwen3.5 4B5 of 5 tests
Following instructionsDoing exactly what it is asked01Qwen3.5 4B1 of 1 test

Following instructions rests on a single test.

What it costs

Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.

LFM2-VL-3BNo current price is tracked.

Qwen3.5 4B$0.03 to read · $0.15 to write$0.18Price from Artificial Analysis

The biggest differences

The tests each model wins by the widest margin, up to three each. Scores are out of 100.

Where Qwen3.5 4B pulls ahead

Harder college exam questions with imagesMMMU-Pro+36.7points aheadQwen3.5 4B65.4LFM2-VL-3B28.7
Follows unfamiliar, precisely checkable instructionsIFBench+31.2points aheadQwen3.5 4B52LFM2-VL-3B20.8
Visual perception tasks people solve at a glanceBLINK+8.5points aheadQwen3.5 4B58.7LFM2-VL-3B50.2

Where LFM2-VL-3B pulls ahead

No clear win on a test scored out of 100.

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 chartsQwen3.5 4B

MMMU-ProQwen3.5 4B by 36.728.765.4+36.7
BLINKQwen3.5 4B by 8.550.258.7+8.5
OCRBenchQwen3.5 4B by 4.782.286.9+4.7
MMStarQwen3.5 4B by 1.657.759.3+1.6
MathVistaQwen3.5 4B by 1.562.163.6+1.5

Full images and charts ranking

Following instructionsQwen3.5 4B

IFBenchQwen3.5 4B by 31.220.852+31.2

Full following instructions ranking

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 WebQwen3.5 4B by 75.32.577.8+75.3
ScreenSpot-v2 MobileQwen3.5 4B by 73.87.681.4+73.8
ScreenSpot-v2 DesktopQwen3.5 4B by 70.3676.3+70.3
BFCLv4Qwen3.5 4B by 30.120.550.6+30.1
OCRBench v2_enQwen3.5 4B by 14.843.958.7+14.8
IFEvalQwen3.5 4B by 13.372.986.2+13.3
InfographicVQA (val)Qwen3.5 4B by 12.567.880.3+12.5
MM-IFEvalQwen3.5 4B by 11.751.463.1+11.7
SimpleVQAQwen3.5 4B by 7.73340.7+7.7
MMEQwen3.5 4B by 6.57379.5+6.5
HallusionBenchQwen3.5 4B by 5.346.451.7+5.3
DocVQA-valQwen3.5 4B by 589.894.8+5
MMMU (val) (Pass@1)Qwen3.5 4B by 4.745.650.3+4.7
RealWorldQALFM2-VL-3B by 471.167.1+4
OCRBench v1Qwen3.5 4B by 3.981.785.6+3.9
ChartQA TestQwen3.5 4B by 3.880.484.2+3.8
POPELFM2-VL-3B by 3.289.286+3.2
TextVQA-valLFM2-VL-3B by 1.88381.2+1.8
MMBenchLFM2-VL-3B by 1.68078.4+1.6
Multilingual MMBenchtie76.377tie
SEED-Bench (image)tie76.676.1tie
MMMBtie81.982tie
Multilingual MMMBtie81.982tie

Questions people ask

Which is better, LFM2-VL-3B or Qwen3.5 4B?

Qwen3.5 4B wins all two areas where both have results: images and charts and following instructions. LFM2-VL-3B wins none.

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.