Benchmark Almanac

LongCat-Flash-Thinking-2601

LongCat-Flash-Thinking-2601 places in the top three on 3 benchmarks.

Meituan Released January 14th, 2026 Family LongCat-Flash The lab's model card Compare with others
20
Benchmarks with a score
0
Ranked first, out of five or more models
3
In the top three
Meituan Jan 2026

LongCat-Flash-Thinking-2601

Open weights Flagship

147.9 #107 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • VITA-Bench 29.3% #3 /18
  • BrowseComp-ZH 77.7% #3 /19
  • RWSearch 79.5% #2 /7

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
Humanity's Last Exam 25.2% #93 of 154 Lab model card

Math and science

BenchmarkScoreRankSource
AIME 2025 100% #8 of 143 Lab model card
HMMT 2025 97.5% #7 of 54 Lab model card
IMO-AnswerBench 86.8% #9 of 37 Lab model card
AMO-Bench 66% #1 of 4 Lab model card
GPQA Diamond 85.2% #76 of 302 Lab model card
General365 50.8% #16 of 23 Official leaderboard

Coding

BenchmarkScoreRankSource
LiveCodeBench v6 82.8% #27 of 90 Lab model card
OJBench 42.2% #5 of 16 Lab model card
SWE-bench Verified 70% #79 of 155 Lab model card
OIBench 47.7% #1 of 1 Lab model card

Agents and tools

BenchmarkScoreRankSource
BrowseComp-ZH 77.7% #3 of 19 Lab model card
VITA-Bench 29.3% #3 of 18 Official leaderboard
tau2-bench 88.2% #18 of 79 Lab model card
RWSearch 79.5% #2 of 7 Official leaderboard
BrowseComp 73.1% #45 of 81 Lab model card
ClawProBench 57.5% #30 of 49 Official leaderboard
Random Complex Tasks (LongCat) 35.8% #1 of 1 Lab model card
tau2-Noise (LongCat) 67.1% #1 of 1 Lab model card
VitaBench-Noise (LongCat) 20.5% #1 of 1 Lab model card

Its lab's numbers put it at 147.9, number 107 of 404

The Self-Reported Index fits one score to the 12 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 145.3 to 150.6. How we compute it.

Its lab's numbers next to 2 leaderboards

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
RWSearch 79.5% RW Search (Pass@1) 79.5% Leaderboard Same
VITA-Bench 29.3% VitaBench (Avg@4) 29.3% Leaderboard, thinking Same

Its launch post

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

LongCat-Flash-Thinking-2601 model card (Hugging Face) Jan 14, 2026

Its system card

Every result in the longer documents the lab published with it.