Mage-VL-4B vs Phi 4 Multimodal Instruct

Microsoft · released Jul 26, 2026

Wins 1 of 1 area

Images and charts

Microsoft · released Feb 26, 2025

Wins 0 of 1 area

—

Mage-VL-4B is the stronger all-rounder.

Scores updated Aug 23, 2026 · 24 tests both models report · How we compare

Where each one wins

Tests won in each of the one area 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 video31Mage-VL-4B3 of 4 tests

The biggest differences

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

Where Mage-VL-4B pulls ahead

Questions about short and long videosVideo-MME+9points aheadMage-VL-4B64Phi 4 Multimodal Instruct55
Questions that truly need the image to answerMMStar+6.1points aheadMage-VL-4B67.3Phi 4 Multimodal Instruct61.2
Visual perception tasks people solve at a glanceBLINK+3.8points aheadMage-VL-4B65.1Phi 4 Multimodal Instruct61.3

Where Phi 4 Multimodal Instruct pulls ahead

Reads text in imagesOCRBench+2.6points aheadPhi 4 Multimodal Instruct84.4Mage-VL-4B81.8

Every test, side by side

All 24 tests both models report. The winning score is in its model's colour; marks a score checked independently.

Images and chartsMage-VL-4B

Video-MMEMage-VL-4B by 96455+9
MMStarMage-VL-4B by 6.167.361.2+6.1
BLINKMage-VL-4B by 3.865.161.3+3.8
OCRBenchPhi 4 Multimodal Instruct by 2.681.884.4+2.6

Full images and charts ranking

Other results20 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.

Timelens-ActivityNetMage-VL-4B by 43.445.42+43.4
MultiDocVQA-valMage-VL-4B by 40.787.546.8+40.7
VSI-BenchMage-VL-4B by 40.264.324.1+40.2
TextVQA-valMage-VL-4B by 37.477.339.9+37.4
MME-RealWorldMage-VL-4B by 3466.532.5+34
CV-BenchMage-VL-4B by 30.787.857.1+30.7
MME-PerceptionMage-VL-4B by 300 rating points17101410+300 rating
NextQAMage-VL-4B by 2983.154.1+29
MLVU-devMage-VL-4B by 24.568.744.2+24.5
Ref-DAVIS17Mage-VL-4B by 22.725.83.1+22.7
LongVideoBenchMage-VL-4B by 20.261.341.1+20.2
MV-BenchMage-VL-4B by 20.265.144.9+20.2
MMBench-EN-devMage-VL-4B by 18.28465.8+18.2
LVBenchMage-VL-4B by 16.541.825.3+16.5
SATMage-VL-4B by 1267.355.3+12
InfoVQA (val)Mage-VL-4B by 8.580.371.8+8.5
MMBench-CN-devMage-VL-4B by 6.88275.2+6.8
ChartQAMage-VL-4B by 3.584.981.4+3.5
DocVQA-valMage-VL-4B by 2.395.192.8+2.3
AI2D w/ MaskMage-VL-4B by 1.483.281.8+1.4

Questions people ask

Which is better, Mage-VL-4B or Phi 4 Multimodal Instruct?

Mage-VL-4B wins the one area where both have results: images and charts. Phi 4 Multimodal Instruct wins none.

How do you compare the two?

We use the 24 benchmark tests both models have published scores on. The verdict counts the 4 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 20 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.