LocalClaw/LLM/LongCat-Flash-Lite
Open-weight MoE

LongCat-Flash-Lite

Catalogue summary: Meituan LongCat open-weight MoE with 68.5B total parameters, 3-4.5B active, MIT licensing and a 256K+ context window. Practical only for 64GB+ workstations through LongCat-specific GGUF/MLX runtimes.

Repository editorial metadata; verify comparative claims in the linked upstream material.

64 GB catalogue minimum64 GB RAMQ4_K_MCoding assistant
Choose an app

This model needs a special runtime. Unsupported apps are clearly marked.

Compare models
LM StudioNot available for this model
UnslothNot available for this model
OllamaNot available for this model
Open on Hugging FaceFiles, licence and available downloads
llama.cppNot available for this model
Use with LocalClawOptional workspace after the model is installed
This model needs its official runtimeLongCat llama.cpp fork / MLX
Parameters
68.5B (3-4.5B active, MoE)
Minimum RAM
64 GB
Model size
37.4 GB
Quantization
Q4_K_M

Can LongCat-Flash-Lite run locally?

LongCat-Flash-Lite has a catalogue minimum of 64 GB RAM with Q4_K_M. Actual memory use and speed vary by context length, runtime, backend and system headroom.

Use the official LongCat llama.cpp fork / MLX setup. The current low-bit files are not a stock LM Studio install.

Model sourcePublic GGUF repository (verified 2026-09-07): InquiringMinds-AI/LongCat-Flash-Lite-GGUFUpstream releasePaper or model cardLicence text
chatcodereasoningagentpowerlong-contextmoe

Install path

01
Check RAM fitMinimum 64 GB RAM. Start with the Q4_K_M quant.
02
Load the modelFollow the official LongCat llama.cpp fork / MLX instructions. Stock LM Studio support is not confirmed.
03
Control locallyUse LocalClaw to manage models, agents, chat, channels and scheduled OpenClaw work.

Catalogue record

  • Family: longcat
  • Parameters: 68.5B (3-4.5B active, MoE)
  • Recommended quantization: Q4_K_M
  • Catalogue minimum RAM: 64 GB
  • Catalogue model size: 37.4 GB
  • Tags: chat, code, reasoning, agent, power, long-context, moe

Practical limits

  • Catalogue RAM is a minimum estimate, not a guarantee for every context length or runtime.
  • Speed and memory use vary by quantization, backend, context length and system headroom.
  • Verify architecture, licence and usage restrictions in the linked upstream material before deployment.

Catalogue tags

  • chat
  • code
  • reasoning
  • agent
  • power
  • long-context

Capability profile

Repository catalogue ratings used by LocalClaw's editorial rubric. They are not a standardized third-party benchmark.

speed
6
quality
8
coding
9
reasoning
8

Technical notes

Developer
Meituan LongCat
License
MIT
Context window
327,680 tokens
Architecture
Sparse MoE language model with 68.5B total parameters, about 3-4.5B active parameters, MLA, identity experts and N-gram embeddings. The official model card describes a 256K context target; the GGUF conversion documents a 327,680-token window.

This model fits these next steps

Hardware fit is based on LocalClaw's RAM tier, model size and quantization metadata. Always leave memory headroom for your OS and runtime.

Workstation fitMac Studio M4 Max 64GBCUDA workstation classNVIDIA GB10 / DGX SparkLarge local models64GB RAM guide

Related catalogue entries

Linked mechanically by family, shared tags and nearby RAM tier; this is not a quality ranking.

Qwen3-Coder-Next 80B (3B active, MoE)Kimi Linear 48B-A3B Instruct 48B (3B active, MoE)Laguna XS 2.1 33B (3B active, MoE)Mistral Small 4 (119B-A6.5B) 119B (6.5B active, MoE)LLM-jp-4 33B Thinking 33BNemotron 3.5 Lightning 30B-A3B 30B (3B active, MoE)Ornith-1.5-35B-A3B 35B (3B active, MoE)Ling Flash Base 2.0 106B (6.1B active, MoE)

Where to go next