HunyuanImage-3.0 (83B-A13B MoE) VRAM & GPU Requirement Calculator

Interactive hardware sizing for HunyuanImage-3.0 (83B-A13B MoE) (83B by Tencent). Estimated 8K context footprint: 51.4 GB (INT4) • 86.2 GB (FP8) • 169.2 GB (FP16).

GPUs

Memory Requirements 51.35 GB

Requires 3 GPUs (based on memory capacity)

48.14 GB

All model weights

0 GB

Conversation history cache

2.31 GB

Expert model optimization

0.9 GB

Temporary computation cache

ComfyUI Pipeline & Hardware Estimates
Est. Gen Time
12.9 s/img
Min System RAM
80 GB RAM
Resolution
1 MP
Offload Mode
Smart Offload
Quick Deploy CLI Command
python main.py --lowvram --preview-method auto
Flags reflect your selected precision, context length, and tensor parallel GPU count.Open HunyuanImage-3.0 spec page →

HunyuanImage-3.0 (83B-A13B MoE) Architecture & Memory Profile

Total Parameters
83B
Architecture
MoE (13B active)
Developer / Provider
Tencent
Recommended Precision
INT4

HunyuanImage-3.0 (83B-A13B MoE) ComfyUI VRAM Matrix by Precision & Resolution

Precomputed 3-stage ComfyUI Smart Offload + VAE Tiling VRAM peak (recommended host RAM: 80 GB).

Precision1024x1024 (1MP)1536x1536 (2.25MP)2048x2048 (4MP)LoRA Train (1MP)
FP16169.2 GB172.6 GB178.4 GB172.5 GB
FP886.2 GB89.6 GB95.4 GB89.5 GB
INT886.2 GB89.6 GB95.4 GB89.5 GB
INT451.4 GB54.8 GB60.5 GB54.6 GB

Recommended GPU Hardware for HunyuanImage-3.0 (83B-A13B MoE) (INT4)

Minimum GPU count for 1024x1024 generation with ComfyUI offload and estimated seconds/image.

GPU HardwareMin GPUsTotal VRAM PoolEst. Gen Time
NVIDIA RTX 4090 (24GB)3×72 GB12.9 s/img
NVIDIA RTX 5090 (32GB)2×64 GB8.8 s/img
Apple M4 Max (128GB) (128GB)1×128 GB29.6 s/img
NVIDIA A100 (80GB) (80GB)1×80 GB25.9 s/img
NVIDIA H100 (80GB)1×80 GB7.1 s/img
NVIDIA H200 (141GB)1×141 GB7 s/img
Huawei Ascend 910B (64GB)1×64 GB13.9 s/img

HunyuanImage-3.0 (83B-A13B MoE) Deployment FAQ

How much VRAM is needed to run HunyuanImage-3.0 (83B-A13B MoE)?
At 1024x1024 with ComfyUI Smart Offload and VAE Tiling enabled, HunyuanImage-3.0 (83B-A13B MoE) needs about 169.2 GB in BF16/FP16, 86.2 GB in FP8, and 51.4 GB in INT4/GGUF (paired with ~80 GB system RAM).
Can I run HunyuanImage-3.0 (83B-A13B MoE) on RTX 4090 or H100 GPUs?
Running HunyuanImage-3.0 (83B-A13B MoE) in INT4 precision requires at least 3x RTX 4090 (24GB) or 1x NVIDIA H100 (80GB) GPUs.
What precision should I choose for HunyuanImage-3.0 (83B-A13B MoE) in production?
For HunyuanImage-3.0 (83B-A13B MoE) (83B by Tencent), INT4 provides the best balance between inference speed, VRAM footprint, and output quality.