What are the hardware requirements for LongCat Flash Thinking?

LongCat Flash Thinking requires approximately 0.1 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed LongCat Flash Thinking and how large is it?

LongCat Flash Thinking is developed by meituan-longcat, a model with 561.9b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does LongCat Flash Thinking perform on standard benchmarks?

LongCat Flash Thinking has the following published scores: LMSYS ELO 1393. Compare against reference models on this page or on the LLM Explorer leaderboards.

LongCat Flash Thinking by meituan-longcat — VRAM 0.1GB, 128K context

Name: LongCat Flash Thinking
Author: meituan-longcat

LongCat Flash Thinking is an open-source language model by meituan-longcat. Features: 561.9b LLM, VRAM: 0.1GB, Context: 128K, License: mit, LLM Explorer Score: 0.36, ELO: 1393.

Arxiv:2509.18883 Conversational Custom code Longcat-flash-chat Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/meituan-longcat/LongCat-Flash-Thinking

LongCat Flash Thinking Benchmarks

LMSys ELO 1393 vs 1272 (so35) 9.5%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

LongCat Flash Thinking Parameters and Internals

LLM Name	LongCat Flash Thinking
Repository 🤗	https://huggingface.co/meituan-longcat/LongCat-Flash-Thinking
Model Size	561.9b
Required VRAM	0.1 GB
Updated	2026-07-06
Maintainer	meituan-longcat
Model Files	1.6 GB: 1-of-75 1.9 GB: 2-of-75 0.2 GB: 3-of-75 1.8 GB: 4-of-75 0.1 GB: 5-of-75 16.1 GB: 6-of-75 16.1 GB: 7-of-75 15.9 GB: 8-of-75 16.0 GB: 9-of-75 15.9 GB: 10-of-75 16.1 GB: 11-of-75 16.1 GB: 12-of-75 15.9 GB: 13-of-75 16.0 GB: 14-of-75 15.9 GB: 15-of-75 16.1 GB: 16-of-75 16.1 GB: 17-of-75 15.9 GB: 18-of-75 16.0 GB: 19-of-75 15.9 GB: 20-of-75 16.1 GB: 21-of-75 16.1 GB: 22-of-75 15.9 GB: 23-of-75 16.0 GB: 24-of-75 15.9 GB: 25-of-75 16.1 GB: 26-of-75 16.1 GB: 27-of-75 15.9 GB: 28-of-75 16.0 GB: 29-of-75 15.9 GB: 30-of-75 16.1 GB: 31-of-75 16.1 GB: 32-of-75 15.9 GB: 33-of-75 16.0 GB: 34-of-75 15.9 GB: 35-of-75 16.1 GB: 36-of-75 16.1 GB: 37-of-75 15.9 GB: 38-of-75 16.0 GB: 39-of-75 15.9 GB: 40-of-75 16.1 GB: 41-of-75 16.1 GB: 42-of-75 15.9 GB: 43-of-75
Model Architecture	LongcatFlashForCausalLM
License	mit
Context Length	131072
Model Max Length	131072
Vocabulary Size	131072

Rank the LongCat Flash Thinking capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to LongCat Flash Thinking

Best Alternatives	Context / RAM	Downloads	Likes
LongCat Flash Chat	128K / 0.1 GB	58246	537
LongCat Flash Chat FP8	128K / 0.1 GB	255	25
LongCat Flash Thinking FP8	128K / 0.1 GB	71	8

Note: green Score (e.g. "73.2") means that the model is better than meituan-longcat/LongCat-Flash-Thinking.