10x more images per dollar.
Image generation is expensive on hard-to-get, AI-focused data-center GPUs. On SaladCloud's RTX 4090s and 5090s the same models deliver far more images per dollar, with the elasticity to serve millions a day.
Published, reproducible, linked.
Flux.1-Schnell
Our Flux.1-Schnell benchmark delivered 5,243 images per dollar across the network, with 4,265 images per dollar on a single RTX 4090. On the RTX 5090, per-image time drops to 1.2 seconds, and TheStage AI's ANNA optimization pushes past 10,000 images per dollar.
Read the benchmark →Stable Diffusion XL
60,600 high-resolution SDXL images generated on RTX 3090 and 4090 GPUs at $0.0013 per image, 3,405 images per dollar.
Read the benchmark →Recipes for the models you already use.
Bring your own: any ComfyUI, Automatic1111, Cog, or custom container runs on Salad Container Engine. Fine-tuning guides for Dreambooth, SDXL LoRA, and ControlNet are on the blog.
Image generation at scale, in production.
"By switching to Salad, Civitai is now serving inference on over 600 consumer GPUs to deliver 10 Million images per day and training more than 15,000 LoRAs per month. Salad not only had the lowest GPU prices in the market but also offered us incredible scalability."
Justin Maier · Founder, CivitaiCase study →"On Salad's consumer GPUs, we are running 3X more scale at half the cost of A100s on our local provider and almost 85% less cost than the two major hyperscalers we were using before. I'm not losing sleep over scaling issues anymore."
Jamsheed Kamardeen · CTO, BlendCase study →"Salad makes it more realistic to keep up with deploying these new models. We might never deploy most of them if we had to pay AWS cost for them."
Shawn Rushefsky · Founder, DreamupCase study →Cut image-generation costs by up to 90%.
Volume pricing for more than 100 GPUs, long-running jobs, and committed contracts.