Best Multimodal LLMs

The benchmarks behind the multimodal ranking

Six public benchmarks decide this ranking. The heavier a benchmark's weight, the more it moves a model's score.

• 21%
  Video-MME

  Questions about short and long videos

  Best on this testSeed 2.1 Pro Preview89.2

• 21%
  OCRBench

  Reads text in images

  Best on this testKimi K2.592.3

• 17%
  MMMU

  College exam questions with charts, maps and diagrams

  Best on this testQwen3.6 Plus86.0

• 15%
  CharXiv (reasoning)

  Reasoning about charts from research papers

  Best on this testClaude Mythos Preview93.2

• 13%
  MMMU-Pro

  Harder college exam questions with images

  Best on this testClaude Opus 5.587.7

• 13%
  MathVista

  Math problems shown in pictures and charts

  Best on this testSeed 2.1 Pro Preview90.7

• Tracked, not ranked

  BLINK — visual perception tasks people solve at a glance. It updates too slowly to score new models within a week, so it sits outside the ranking; its scores appear on each model's page, and it returns to the ranking when it keeps pace.

• Tracked, not ranked

  MMStar — questions that truly need the image to answer. It updates too slowly to score new models within a week, so it sits outside the ranking; its scores appear on each model's page, and it returns to the ranking when it keeps pace.

• Tracked, not ranked

  OCRBench v2 — reads and reasons about text in images. It updates too slowly to score new models within a week, so it sits outside the ranking; its scores appear on each model's page, and it returns to the ranking when it keeps pace.

• Tracked, not ranked

  LMArena Vision — people's blind votes on answers about images. It updates too slowly to score new models within a week, so it sits outside the ranking; its scores appear on each model's page, and it returns to the ranking when it keeps pace.

How the multimodal score is calculated

1. 1

   Rank on each benchmark

   Every model gets a percentile on each benchmark it has been measured on.

2. 2

   Steady the thin fields

   Where few models have taken a benchmark, that percentile is pulled toward the middle of the field.

3. 3

   Weigh and average

   The percentiles are averaged with the weights above into one score out of 100.

4. 4

   Qualify

   A model enters once it is measured on at least half the basket by weight, including one anchor benchmark.

190 models from 30 vendors have a multimodal score, each measured on at least 3 of the 6 benchmarks. Scores come from AA-graded results, official model cards and third-party evaluations.

Questions about the multimodal ranking