Atlas

Models

← All models

LongCat-Flash-Lite

Meituan · 2026-02-05 · 69.1B parameters

LongCat-Flash-Lite is Meituan's non-thinking sparse mixture-of-experts language model, published as a 68.5B-parameter model with 2.9B to 4.5B parameters dynamically activated per token. Trained from scratch, it allocates 31.4B parameters to an N-gram embedding module rather than additional experts to improve model performance and inference efficiency. It supports contexts up to 256K tokens using YaRN and targets agentic tool use and coding.

Benchmark scores

BenchmarkScore
AA-LCR25.7
AA-Omniscience Index-70.1
Artificial Analysis Intelligence Index17.2
Artificial Analysis Omniscience Accuracy13.1
Artificial Analysis Omniscience Hallucination Rate95.7
Artificial Analysis Openness Index44.4
Capability133.9
CritPt0.0
GPQA Diamond63.6
Humanity's Last Exam (Text-Only)6.0
IFBench (Artificial Analysis)43.1
SciCode (Artificial Analysis)28.4
Terminal-Bench Hard10.6
τ²-bench (Telecom)79.5
Loading Atlas data…