Home/FAQ & Glossary/Can I run local image generation (Flux / SDXL) and LLMs on the same rig?
FREQUENTLY ASKED QUESTIONconcepts
← Back to FAQ & Glossary

Can I run local image generation (Flux / SDXL) and LLMs on the same rig?

Yes. The exact same hardware built for local LLMs will also handle image generators like Flux and Stable Diffusion XL (SDXL) well. If your machine already has enough VRAM for a solid text model, it already meets the baseline for local image generation.

That said, running both workloads on one PC comes down to three practical differences:

1. Taking turns vs. running both at the exact same time

  • One after the other (Works on 12GB–16GB+ VRAM): You can easily run Ollama or LM Studio for chat alongside ComfyUI or Forge for images on the same machine—just make sure the text model unloads from VRAM before you generate an image. Ollama automatically unloads idle models after 5 minutes by default, or immediately when you run ollama stop. If an 8B or 32B LLM is still sitting in VRAM when you hit "Generate" in ComfyUI, your GPU will run out of memory and crash.
  • Simultaneously (Needs 24GB–32GB+ VRAM): If you want an AI agent to draft image prompts and immediately render them without swapping models back and forth, you need enough VRAM to hold both at once. A 24GB–32GB card can comfortably keep an 8B–14B LLM (~6–10GB) and Flux.1 Dev FP8 (~10–12GB) loaded side-by-side.

2. How much VRAM do Flux and SDXL actually need?

Unlike LLMs—which are gated mainly by memory bandwidth—image models lean heavily on raw GPU compute (Tensor Cores) and model precision:

Image ModelMinimum VRAMComfortable VRAMGeneration Speed (RTX 3090 / 4090)
SDXL (Base + LoRAs)8 GB (FP16)12 GB (fits ControlNet & refiners)~3 – 5 seconds per image
FLUX.1 Schnell / Dev (FP8 / NF4)8 – 10 GB12 – 16 GB~3 – 6 seconds (Schnell) / ~15 – 30s (Dev)
FLUX.1 Dev (Full FP16 + ControlNet)16 GB24 GB+~15 – 30 seconds per image

Flux also pairs its image generator with a large text encoder called T5-XXL, which is why having 32GB+ of system RAM is strongly recommended so ComfyUI can swap pieces of Flux smoothly without stalling.

3. NVIDIA matters much more for images than for text

While AMD Radeon cards and Apple Silicon Macs do a respectable job running text LLMs, NVIDIA RTX GPUs are dramatically faster at image generation. Popular tools like ComfyUI, Automatic1111, and Stable Diffusion WebUI Forge are built around NVIDIA's CUDA and Tensor Cores—often rendering Flux or SDXL images 2x to 4x faster than a similarly priced Mac or AMD card.

If you plan to use one rig for both local chat/coding and image generation, a 16GB or 24GB NVIDIA GPU (such as a used RTX 3090 or RTX 5070 Ti) paired with 32GB–64GB of system RAM is the most versatile all-in-one setup you can buy.

Looking for verified hardware rigs?

Check our real-time £/GB VRAM index to find systems optimized for local model inference.

Explore VRAM Index →
Rent Cloud GPU Power

Not Ready to Buy a Full Rig?

Don't want to spend £3,500+ upfront? Rent the GPU power instead.

Buying a dedicated AI workstation is a serious investment—and if you aren't ready to spend that much right now, you don't have to. Renting cloud GPUs gives you instant access to high-VRAM hardware like RTX 4090s and H100s on demand. Spin up an instance for fine-tuning or running massive 70B+ models, pay strictly for the minutes you use, and avoid the upfront hardware cost, electricity bills, and maintenance.

Vast.ai Global GPU Marketplace

Lowest $/hr
Unbeatable $/hr rates on community-hosted RTX 4090 & 3090 rigs
Deploy Pod →

RunPod Community & Secure Cloud

Community & Secure Cloud
RTX 4090 from $0.34/hr • H100 SXM from $1.99/hr (Billed in USD)
Deploy Pod →

FindBestRig independently benchmarks cloud GPU providers to compare rental economics against physical hardware.