Benchmark Almanac

LongCat-Flash-Omni

LongCat-Flash-Omni places in the top three on 3 benchmarks.

Meituan Released October 31st, 2025 Family LongCat-Flash The lab's model card Compare with others
54
Benchmarks with a score
0
Ranked first, out of five or more models
3
In the top three
Meituan Oct 2025

LongCat-Flash-Omni

Open weights Flagship

132.9 #250 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • MMVU 67.1% #20 /26
  • MVBench 75.2% #7 /23
  • BLINK 63.1% #16 /29

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
MMLU 90.3% #12 of 131 Lab model card
C-Eval 91.7% #11 of 46 Lab model card
MMLU-Pro 82.7% #51 of 167 Lab model card
CMMLU 89.4% #7 of 17 Lab model card
DROP 83.5% #17 of 41 Lab model card
IFEval 82.4% #96 of 135 Lab model card
GraphWalks 56% #23 of 27 Lab model card
COLLIE 45.7% #11 of 12 Lab model card
Meeseeks 39% #2 of 2 Lab model card

Math and science

BenchmarkScoreRankSource
MATH-500 97.6% #12 of 70 Lab model card
GPQA Diamond 74.4% #153 of 302 Lab model card
ZebraLogic 86% #12 of 23 Lab model card
AIME 2024 72.9% #56 of 88 Lab model card
Beyond AIME 47.4% #13 of 14 Lab model card

Coding

BenchmarkScoreRankSource
MBPP+ 80.2% #5 of 15 Lab model card
HumanEval+ 90.8% #10 of 27 Lab model card
LiveCodeBench 52.6% #27 of 54 Lab model card
LiveCodeBench v6 52.6% #79 of 90 Lab model card

Agents and tools

BenchmarkScoreRankSource
AndroidControl 91.2% #1 of 2 Lab model card
ScreenSpot-v2 91.2% #10 of 14 Lab model card
VisualWebBench 78.7% #2 of 2 Lab model card

Vision, audio and robotics

BenchmarkScoreRankSource
RefCOCO / RefCOCO+ / RefCOCOg 93.9% #2 of 29 Lab model card
MVBench 75.2% #7 of 23 Lab model card
MuirBench 77.1% #4 of 12 Lab model card
MMBench 87.5% #4 of 11 Lab model card
ChartQA 87.6% #14 of 37 Lab model card
UNO-Bench 46.3% #3 of 7 Official leaderboard
MathVista 77.9% #31 of 70 Lab model card
VoiceBench 88.7% #6 of 12 Lab model card
MM-Vet 69% #4 of 8 Lab model card
WorldSense (omnimodal) 60.9% #3 of 6 Lab model card
Mantis-Eval 84.8% #1 of 2 Lab model card
OmniBench 61.4% #1 of 2 Lab model card
OpenAudioBench 80.1% #1 of 2 Lab model card
MMAU 75.9% #8 of 15 Lab model card
BLINK 63.1% #16 of 29 Lab model card
CountBench 92.4% #16 of 28 Lab model card
Daily-Omni 82.4% #4 of 7 Lab model card
TempCompass 82.2% #4 of 7 Lab model card
MMMU 70.7% #60 of 103 Lab model card
RealWorldQA 74.8% #23 of 38 Lab model card
Video-MME 78.2% #34 of 51 Lab model card
Clotho-AQA 72.8% #2 of 3 Lab model card
DocVQA 91.8% #25 of 36 Lab model card
CoVoST2 37.3 #10 of 14 Lab model card
LongVideoBench 69.3% #8 of 11 Lab model card
OCRBench 84.9% #25 of 34 Lab model card
MMStar 70.9% #27 of 36 Lab model card
AISHELL-1 0.63% #3 of 4 Lab model card
MMVU 67.1% #20 of 26 Lab model card
Video-MMMU 67.5% #30 of 36 Lab model card
AISHELL-2 2.78% #3 of 3 Lab model card
NExT-QA 86.2% #1 of 1 Lab model card
Real-time Audio-Visual Interaction Evaluation 1.37 #1 of 1 Lab model card

Its lab's numbers put it at 132.9, number 250 of 404

The Self-Reported Index fits one score to the 36 capability results Meituan reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 131.2 to 134.6. How we compute it.

Its lab's number next to a leaderboard

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark Lab says Board says Gap
UNO-Bench 49.9% UNO-Bench 46.3% Leaderboard +3.6 pts

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

LongCat-Flash-Omni model card (Hugging Face) Oct 31, 2025

Its system card

Every result in the longer documents the lab published with it. 2 of them aren't in the launch post.

LongCat-Flash-Omni Technical Report Oct 31, 2025

Capabilities