CodeGeeX4 9B

THUDM/codegeex4-all-9b

Parameters

9.4B

Context

128K

Downloads

9.5K

Likes

272

Architecture

transformer

View on HuggingFace

GPU Requirements

VRAM / RAM / disk estimates by quantization and usage scenario

QuantizationMinimumRecommendedProduction
FP1616bit
27.1 GB
RAM 40.7 · Disk 66.3
43.9 GB
RAM 65.8 · Disk 92.3
141.8 GB
RAM 212.6 · Disk 131.3
FP88bit
16.0 GB
RAM 32.0 · Disk 52.7
32.3 GB
RAM 48.4 · Disk 78.7
129.7 GB
RAM 194.5 · Disk 117.7
INT88bit
16.0 GB
RAM 32.0 · Disk 52.7
32.3 GB
RAM 48.4 · Disk 78.7
129.7 GB
RAM 194.5 · Disk 117.7
AWQ4bit
10.7 GB
RAM 32.0 · Disk 46.1
26.7 GB
RAM 40.1 · Disk 72.1
123.9 GB
RAM 185.8 · Disk 111.1
GPTQ4bit
10.8 GB
RAM 32.0 · Disk 46.2
26.8 GB
RAM 40.2 · Disk 72.2
123.9 GB
RAM 185.9 · Disk 111.2
GGUF4bit
10.9 GB
RAM 32.0 · Disk 46.3
26.9 GB
RAM 40.4 · Disk 72.3
124.1 GB
RAM 186.1 · Disk 111.3

GPU Recommendations

Best ValueBEST

1× H100 SXM

80 GB VRAM

$1,170/mo

≈ 737.9 tok/s

Effective production throughput ≈ 516.5 tok/s

score 0.5786

Deploy on vast ↗
Cheapest

2× Tesla V100

32 GB VRAM

$40.6/mo

≈ 0 tok/s (N/A — benchmark unavailable)

Effective production throughput ≈ 0 tok/s

score 0.6241

Deploy on vast ↗
Performance

8× H200 SXM

1128 GB VRAM

$20,440/mo

≈ 3383.3 tok/s

Effective production throughput ≈ 2368.3 tok/s

score 0.2197

Min Complexity

1× RTX A6000

48 GB VRAM

$210/mo

≈ 169.2 tok/s

Effective production throughput ≈ 118.4 tok/s

score 0.6395

Deploy on vast ↗

💬 Community — real deployment experience

0 Articles · 0 Benchmarks

View Community →

GPUs that fit (single card)

RTX A6000

48 GB VRAM

$210

/mo · vast

RTX 5090

32 GB VRAM

$236

/mo · vast

L40S

48 GB VRAM

$341

/mo · vast

RTX PRO 6000 WS

48 GB VRAM

$482

/mo · vast

RTX PRO 6000 S

48 GB VRAM

$633

/mo · vast

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