Coding AIGLMMIT

GLM-5.2

zai-org/GLM-5.2

Parameters

744B

Context

1024K

Downloads

2.5M

Likes

4.9K

Architecture

moe_transformer

View on HuggingFace

GPU Requirements

VRAM / RAM / disk estimates by quantization and usage scenario

QuantizationMinimumRecommendedProduction
FP1616bit
1758.0 GB
RAM 2637.0 · Disk 2200.9
1853.4 GB
RAM 2780.2 · Disk 2226.9
2030.0 GB
RAM 3045.0 · Disk 2265.9
FP88bit
881.5 GB
RAM 1322.2 · Disk 1119.9
937.1 GB
RAM 1405.6 · Disk 1145.9
1073.8 GB
RAM 1610.7 · Disk 1184.9
INT88bit
881.5 GB
RAM 1322.2 · Disk 1119.9
937.1 GB
RAM 1405.6 · Disk 1145.9
1073.8 GB
RAM 1610.7 · Disk 1184.9
AWQ4bit
459.7 GB
RAM 689.5 · Disk 599.7
496.1 GB
RAM 744.1 · Disk 625.7
613.6 GB
RAM 920.5 · Disk 664.7
GPTQ4bit
465.1 GB
RAM 697.7 · Disk 606.5
501.8 GB
RAM 752.7 · Disk 632.5
619.6 GB
RAM 929.4 · Disk 671.5
GGUF4bit
476.1 GB
RAM 714.1 · Disk 620.0
513.3 GB
RAM 769.9 · Disk 646.0
631.6 GB
RAM 947.3 · Disk 685.0

GPU Recommendations

Best ValueBEST

8× H200 SXM

1128 GB VRAM

$20,440/mo

≈ 42.7 tok/s

Effective production throughput ≈ 29.9 tok/s

score 0.337

Cheapest

8× A100 80GB

640 GB VRAM

$7,008/mo

≈ 18.2 tok/s

Effective production throughput ≈ 12.7 tok/s

score 0.4539

Deploy on vast ↗
Performance

4× H200 SXM

564 GB VRAM

$10,220/mo

≈ 32 tok/s

Effective production throughput ≈ 22.4 tok/s

score 0.4348

Min Complexity

4× H200 SXM

564 GB VRAM

$10,220/mo

≈ 32 tok/s

Effective production throughput ≈ 22.4 tok/s

score 0.4348

💬 Community — real deployment experience

0 Articles · 0 Benchmarks

View Community →
AmciHub

AI Model Deployment & Compute Intelligence — analyze AI model requirements, GPU performance, cloud pricing and deployment costs to find the right deployment solution.

© 2026 AmciHub. All rights reserved. Model → Requirement → GPU → Benchmark → Cloud → Cost → Recommendation