Skip to content
English

Switch theme

// Meituan LongCat · Text LLM

LongCat-Flash-Thinking-ZigZag: VRAM & local deployment

561.9B (MoE) parameters. About 376 GB at 4-bit and 1.24 TB at BF16 with one 8K-token request.

released 2026-01-23licence: Commercial OKmemory figures are estimates

01Specs

PublisherMeituan · meituan-longcat
Family / seriesMeituan LongCat · LongCat-Flash-Thinking-ZigZag
TypeText LLM (text-generation)
Parameters561.9B (MoE) Source: counted from official safetensors weights
Context length960K tokens (983,040)max_position_embeddings in config.json
Architecture28 layers · MLA · MoE
Released weightsBF16 · 1.12 TBdownload
LicenceMIT · Commercial OKPermissive licence (Apache-2.0 / MIT / CC-BY / OpenMDW): commercial use allowed.
Released2026-01-23 (Hugging Face repo created)

02How much VRAM it needs (estimate)

Total = weights + 10% + 0.5GB overhead + KV cache for one request (FP16 KV). More concurrent users multiply the KV column.

PrecisionWeightsTotal @ 8KTotal @ 32KTotal @ 128KSmallest GPUMac
BF16 / FP16 (original)1.12 TB 1.24 TB1.24 TB1.24 TB 8 × B200 (1,440GB)—
FP8 / INT8 / Q8_0597 GB 657 GB658 GB661 GB 8 × H200 (1,128GB)—
4-bit (Q4_K_M / AWQ / GPTQ-Int4)341 GB 375 GB376 GB379 GB 8 × H100 (640GB)—

KV cache per request: 8K 0.3 GB · 32K 1.1 GB · 128K 4.2 GB

Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →

03How to run it

Only runtimes the publisher or the runtime’s official library names for this model.

Which inference engine to choose →

04Official repositories & quantised builds

1 official repo(s) from meituan-longcat.

05Other Meituan LongCat models

All 16 →
ModelParams4-bit totalReleased
LongCat-Flash-Lite-Sparse69B / 3B46.6 GB2026-07-31
LongCat-2.01.6T / 48B1.07 TB2026-07-05
LongCat-AudioDiT-1B1.42B1.4 GB2026-03-30
LongCat-AudioDiT-3.5B3.83B3.1 GB2026-03-30
LongCat-Image-Edit-Turbo——2026-02-03
LongCat-Flash-Lite68.5B / 3B46.3 GB2026-01-27
LongCat-Flash-Thinking-2601560B / 27B374 GB2026-01-14
LongCat-HeavyMode-Summary561.9B376 GB2026-01-14

06Sources

Data checked 2026-02-02. Parameters, context and licence are copied from the official repository; memory figures are our estimates.

Data updated: Self-host open models(UTC+8)