MiniMax

Comfy-Org/MiniMax-H3

Comfy-Org/MiniMax-H3

Parameters

33.1B

Context

Downloads

10.4M

Likes

1.3K

Architecture

View on HuggingFace

GPU Requirements

VRAM / RAM / disk estimates by quantization and usage scenario

QuantizationMinimumRecommendedProduction
FP1616bit
83.0 GB
RAM 124.5 · Disk 135.3
102.3 GB
RAM 153.4 · Disk 161.3
202.7 GB
RAM 304.1 · Disk 200.3
FP88bit
44.0 GB
RAM 66.0 · Disk 87.1
61.5 GB
RAM 92.3 · Disk 113.1
160.2 GB
RAM 240.3 · Disk 152.1
INT88bit
44.0 GB
RAM 66.0 · Disk 87.1
61.5 GB
RAM 92.3 · Disk 113.1
160.2 GB
RAM 240.3 · Disk 152.1
AWQ4bit
25.2 GB
RAM 37.8 · Disk 64.0
41.9 GB
RAM 62.8 · Disk 90.0
139.7 GB
RAM 209.5 · Disk 129.0
GPTQ4bit
25.4 GB
RAM 38.2 · Disk 64.3
42.1 GB
RAM 63.2 · Disk 90.3
140.0 GB
RAM 209.9 · Disk 129.3
GGUF4bit
25.9 GB
RAM 38.9 · Disk 64.9
42.6 GB
RAM 63.9 · Disk 90.9
140.5 GB
RAM 210.7 · Disk 129.9

GPU Recommendations

Best ValueBEST

1× H100 SXM

80 GB VRAM

$1,178/mo

≈ 209.4 tok/s

Effective production throughput ≈ 146.6 tok/s

score 0.6289

Deploy on vast ↗
Cheapest

2× RTX 3090

48 GB VRAM

$186/mo

≈ 93.6 tok/s

Effective production throughput ≈ 65.5 tok/s

score 0.6218

Deploy on vast ↗
Performance

8× H200 SXM

1128 GB VRAM

$20,440/mo

≈ 960 tok/s

Effective production throughput ≈ 672 tok/s

score 0.2197

Min Complexity

1× RTX A6000

48 GB VRAM

$215/mo

≈ 48 tok/s

Effective production throughput ≈ 33.6 tok/s

score 0.6584

Deploy on vast ↗

💬 Community — real deployment experience

0 Articles · 0 Benchmarks

View Community →

GPUs that fit (single card)

RTX A6000

48 GB VRAM

$215

/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

$634

/mo · vast

A100 80GB

80 GB VRAM

$876

/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