LongCat Flash Lite

LongCat

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$0 / $0

in / out

Context

—

Avg score

36.7

7 benchmarks

AA Intelligence

17.2

index

Weights

Closed

Released

Jan 28, 2026

About

LongCat Flash Lite is a proprietary model in the LongCat family from LongCat. Benchmarked on 7 evals, averaging 36.7.

Benchmark scores · 7

avg 36.7 — higher is better

τ²-benchartificial-analysis · 2026-07

79.5

GPQA Diamondartificial-analysis · 2026-07

63.6

IFBenchartificial-analysis · 2026-07

43.1

SciCodeartificial-analysis · 2026-07

28.4

AA Long-Context Reasoningartificial-analysis · 2026-07

25.7

Terminal-Bench Hardartificial-analysis · 2026-07

10.6

Humanity's Last Examartificial-analysis · 2026-07

The research behind this model

all papers mentioning it →

LongCat-Flash Technical Report1 cites · Sep 1, 2025 Multi-Objective and Mixed-Reward Reinforcement Learning via Reward-Decorrelated Policy OptimizationMay 13, 2026

Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.

Details

PublisherLongCat

FamilyLongCat

Parameters—

License—

WeightsClosed

ReleasedJan 28, 2026

History

full ledger →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0/M