Best larger-system route
Team 32
The smallest system in the range that reaches this model's preferred working allowance.
Explore Team 32Stability AI model
Stable Diffusion 3.5 Hardware Requirements & Fit
The repository contains several components and representations, so its 71.6GB file total is not one loaded model. A 24GB offload or optimised workflow is plausible, while 32GB is the entry product envelope and 96GB provides room for less offload, larger workflows and companion models.
- Model version
stabilityai/stable-diffusion-3.5-large- Family and variant
- Stable Diffusion 3.5 Large · stable-diffusion-3.5-large
- Source version
ceddf0a7fdf2- Source updated
- 22 October 2024
Stability AI Stable Diffusion 3.5 Large
stabilityai/stable-diffusion-3.5-large - Minimum GPU memory
- 24GB planning floor with an optimised or offload workflow
- Recommended hardware
- One 32GB GPU for an entry production-style worker
- Licence
- Stability AI Community Licence
- Useful for
- Image generation
Buyer verdict
Where Stable Diffusion 3.5 Large is a sensible fit
Stable Diffusion 3.5 Large remains relevant for teams invested in the Stable Diffusion ecosystem, LoRA workflows and local creative tooling.
These figures apply to the named model version. Quantisation, fine-tuning, context length, image resolution, batch size and serving software can materially change the hardware needed.
Hardware requirements
- Minimum
- 24GB planning floor with an optimised or offload workflow
- Recommended
- One 32GB GPU for an entry production-style worker
The exact pipeline and component placement must be recorded.
Larger resolutions, batches, ControlNets or training can justify 96GB.
Technical specification
Stable Diffusion 3.5 Large model and hardware facts
Specifications shown for source version
ceddf0a7fdf2064ea28e2213e3b84e4afa170a0f, updated
22 October 2024.
- Parameters
- About 8B transformer
- Task
- Text-to-image
- Repository
- Multiple model components and formats
- Libraries
- Diffusers + ComfyUI ecosystem
- Access
- Licence acceptance required
- Licence
- Stability AI Community Licence
Product compatibility
Stable Diffusion 3.5 Large compatibility across all 11 GPU systems
- Systems that fit without model splitting
- 11
- Systems needing multi-GPU validation
- 0
A system is listed as fitting when its GPU memory meets the requirement shown above. Speed, usable context, batch size and concurrent users still need testing with the final model and software configuration.
sme workstation
Team 32
- Per GPU
- 32GB
- Total GPU memory
- 32GB
- GPU count
- 1
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
sme workstation
Company 64
- Per GPU
- 32GB
- Total GPU memory
- 64GB
- GPU count
- 2
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
sme workstation
Studio 96
- Per GPU
- 96GB
- Total GPU memory
- 96GB
- GPU count
- 1
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
sme workstation
Studio 192
- Per GPU
- 96GB
- Total GPU memory
- 192GB
- GPU count
- 2
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
pcie rack
Value Rack 128
- Per GPU
- 32GB
- Total GPU memory
- 128GB
- GPU count
- 4
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
pcie rack
Value Rack 256
- Per GPU
- 32GB
- Total GPU memory
- 256GB
- GPU count
- 8
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
pcie rack
Enterprise 384
- Per GPU
- 96GB
- Total GPU memory
- 384GB
- GPU count
- 4
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
pcie rack
Enterprise 768
- Per GPU
- 96GB
- Total GPU memory
- 768GB
- GPU count
- 8
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
pcie rack
H200 1.1TB
- Per GPU
- 141GB
- Total GPU memory
- 1,128GB
- GPU count
- 8
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
frontier partner
Frontier Native 2.3TB
- Per GPU
- 288GB
- Total GPU memory
- 2,304GB
- GPU count
- 8
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
frontier partner
Frontier Rack 20TB
- Per GPU
- 288GB
- Total GPU memory
- 20,736GB
- GPU count
- 72
Also meets this model page's recommended working allowance.
Available GPU memory exceeds the calculated or estimated requirement. Confirm the final precision, software, workload and performance before purchase.
Deployment reality
Strengths, limits and runtime route
What it is good at
- Large community and workflow ecosystem.
- Diffusers and ComfyUI support.
- Suitable for text-to-image, adapters and controlled fine-tuning workflows.
- Multiple quantised and optimised execution routes exist.
Where to be cautious
- The Stability AI Community Licence has revenue and use conditions and is not Apache or MIT.
- Repository total includes multiple components and should not be treated as loaded VRAM.
- Community nodes and quantisations vary in provenance and quality.
- Training, LoRA work and inference have different memory requirements.
Serving software
- Diffusers
- ComfyUI
- Stability reference workflows
Before installation
- Choose the official checkpoint plus one pinned workflow before sizing.
- Record resolution, steps, guidance, batch, adapters, latency and peak VRAM.
- Review commercial eligibility under the current Stability licence.
- Test generation and LoRA training separately.
System requirements
- GPU layout
- The minimum can fit on one GPU; multiple GPUs may still be useful for replicas or throughput.
- Context and cache
- The model context ceiling is not a guaranteed serving target. KV cache, batch size and concurrent sessions need separate capacity tests.
- System RAM
- Size system memory for model loading, runtime overhead, preprocessing and any CPU offload used by the final configuration.
- Storage
- Allow space for the pinned checkpoint, runtime images, caches, logs and at least one rollback version.
- Serving software
- Validate the exact checkpoint with Diffusers, ComfyUI, Stability reference workflows before acceptance.
- Representative workload
- Benchmark representative prompts or media at the required context, quality, latency and concurrency.
Performance: speed, usable context and concurrency depend on the selected system, software and workload. No benchmark is quoted on this page.
Commercial and legal boundary
Stability AI Community Licence
Commercial use: conditional
Review the current revenue, registration and acceptable-use conditions before commercial deployment.
Always retain the applicable notices and recheck the live terms for the intended organisation, territory, use and distribution route. Obtain legal advice where required; the official licence governs use.
Read the official licenceTechnical questions
Stable Diffusion 3.5 Large deployment FAQ
Can Team 32 run Stable Diffusion 3.5 Large?
It is a plausible 32GB worker for an optimised workflow. The exact pipeline, resolution and peak VRAM need testing.
Is Stable Diffusion 3.5 free for commercial use?
Use is governed by the Stability AI Community Licence, which has conditions. Review the current terms for the organisation and revenue profile.
Do I need 96GB?
Not for every inference workflow. It becomes useful for less offload, larger workflows, multiple components or training work.
What a complete Stable Diffusion 3.5 Large deployment needs
GPU memory is only one part of the system. Storage, data access, serving software, monitoring and administrator handover also affect a reliable deployment.
