image workload · 30 runs on record
SDXL 1024x1024
Stable Diffusion XL 1.0 base, 30 steps, 1024x1024, FP16.
Primary metric: Images / min (img/min)
Reference prompts
Representative prompts for this workload. Exact prompts and harness settings still depend on the cited source for each record.
- Prompt 1
A photorealistic image of a fox playing chess in a library, cinematic lighting, 8k.
- Prompt 2
A flat vector logo of a cyber-tiger holding a microchip, neon palette.
- Prompt 3
An isometric voxel render of a small AI homelab in a closet, cables visible.
Reference runtime command
A representative invocation for reproducing this workload class. Source-specific runs may use adjacent runtimes unless the record says otherwise.
shell
python sdxl_bench.py --model stabilityai/stable-diffusion-xl-base-1.0 --steps 30 --resolution 1024 --seed 42 --batch 1SDXL 1.0 base, 30 steps DPM++ 2M Karras, FP16. We report images/min (sustained, 100-image run).
Full leaderboard
Every record for SDXL image gen.
Top 10 leaderboard
SDXL 1024x1024 · sorted by images / min
30 entries
Value frontier, MSRP vs images / min
Each dot is a device. Top-left is best value (cheap + fast).
Full leaderboard
Click any row for detailed breakdown. Click column headers to sort.
| # | Device | Verif | Buy | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | NVIDIA B200 192GBDatacenter GPU NVIDIA·FP16 | 6 | 52.0img/min | 192 GB | 1000 W | $40k | — | — | Amazon | |
| 2 | NVIDIA GeForce RTX 5090 32GBConsumer GPU NVIDIA·FP16 | 6 | 38.0img/min | 32 GB | 575 W | $2.0k | — | — | Amazon | |
| 3 | NVIDIA H200 141GBDatacenter GPU NVIDIA·FP16 | 6 | 36.0img/min | 141 GB | 700 W | $30k | — | — | Amazon | |
| #4 | NVIDIA GeForce RTX 5090 32GBConsumer GPU NVIDIA·FP16 | 6 | 32.0img/min | 32 GB | 575 W | $2.0k | — | — | Amazon | |
| #5 | NVIDIA H100 SXM5 80GBDatacenter GPU NVIDIA·FP16 | 6 | 28.0img/min | 80 GB | 700 W | $25k | — | — | Amazon | |
| #6 | NVIDIA GeForce RTX 4090 24GBConsumer GPU NVIDIA·FP16 | 6 | 24.0img/min | 24 GB | 450 W | $1.6k | — | — | Amazon | |
| #7 | AMD Instinct MI300X 192GBDatacenter GPU AMD·FP16 | 6 | 22.0img/min | 192 GB | 750 W | $18k | — | — | Amazon | |
| #8 | NVIDIA RTX 6000 Ada 48GBPro GPU NVIDIA·FP16 | 6 | 22.0img/min | 48 GB | 300 W | $6.8k | — | — | Amazon | |
| #9 | AMD Instinct MI300X 192GBDatacenter GPU AMD·FP16 | 6 | 22.0img/min | 192 GB | 750 W | $15k | — | — | Amazon | |
| #10 | NVIDIA L40S 48GBDatacenter GPU NVIDIA·FP16 | 6 | 18.0img/min | 48 GB | 350 W | $7.8k | — | — | Amazon | |
| #11 | NVIDIA GeForce RTX 5080 16GBConsumer GPU NVIDIA·FP16 | 6 | 18.0img/min | 16 GB | 360 W | $999 | — | — | Amazon | |
| #12 | NVIDIA GeForce RTX 5090 32GBConsumer GPU NVIDIA·FP16 | 6 | 18.0img/min | 32 GB | 575 W | $2.0k | — | — | Amazon | |
| #13 | NVIDIA GeForce RTX 4080 Super 16GBConsumer GPU NVIDIA·FP16 | 6 | 16.0img/min | 16 GB | 320 W | $999 | — | — | Amazon | |
| #14 | NVIDIA GeForce RTX 5070 Ti 16GBConsumer GPU NVIDIA·FP16 | 6 | 14.0img/min | 16 GB | 300 W | $749 | — | — | Amazon | |
| #15 | AMD Radeon RX 7900 XTX 24GBConsumer GPU AMD·FP16 | 6 | 13.0img/min | 24 GB | 355 W | $999 | — | — | Amazon | |
| #16 | NVIDIA GeForce RTX 3090 24GBConsumer GPU NVIDIA·FP16 | 6 | 12.0img/min | 24 GB | 350 W | $1.5k | — | — | Amazon | |
| #17 | NVIDIA GeForce RTX 4070 Ti 12GBConsumer GPU NVIDIA·FP16 | 6 | 11.0img/min | 12 GB | 285 W | $799 | — | — | Amazon | |
| #18 | NVIDIA GeForce RTX 5070 12GBConsumer GPU NVIDIA·FP16 | 6 | 10.5img/min | 12 GB | 250 W | $549 | — | — | Amazon | |
| #19 | NVIDIA GeForce RTX 4070 12GBConsumer GPU NVIDIA·FP16 | 6 | 9.0img/min | 12 GB | 200 W | $599 | — | — | Amazon | |
| #20 | NVIDIA GeForce RTX 3090 24GBConsumer GPU NVIDIA·FP16 | 6 | 7.5img/min | 24 GB | 350 W | $1.5k | — | — | Amazon | |
| #21 | NVIDIA GeForce RTX 4060 Ti 16GBConsumer GPU NVIDIA·FP16 | 6 | 6.5img/min | 16 GB | 165 W | $499 | — | — | Amazon | |
| #22 | AMD Radeon RX 7900 XTX 24GBConsumer GPU AMD·FP16 | 6 | 6.5img/min | 24 GB | 355 W | $999 | — | — | Amazon | |
| #23 | Intel Arc B580 12GBConsumer GPU Intel·FP16 | 6 | 6.2img/min | 12 GB | 190 W | $249 | — | — | Amazon | |
| #24 | Intel Arc B580 12GBConsumer GPU Intel·FP16 | 6 | 6.0img/min | 12 GB | 190 W | $249 | — | — | Amazon | |
| #25 | Apple M4 Max (40c GPU, 128GB)Apple Silicon Apple·FP16 | 6 | 6.0img/min | 128 GB | 65 W | $4.7k | — | — | Amazon | |
| #26 | Intel Arc A770 16GBConsumer GPU Intel·FP16 | 6 | 5.0img/min | 16 GB | 225 W | $329 | — | — | Amazon | |
| #27 | Apple M4 Max (40c GPU, 128GB)Apple Silicon Apple·FP16 | 6 | 4.5img/min | 128 GB | 70 W | $4.7k | — | — | Amazon | |
| #28 | Apple M3 Ultra (80c GPU, 512GB)Apple Silicon Apple·FP16 | 6 | 4.2img/min | 512 GB | 80 W | $8.5k | — | — | Amazon | |
| #29 | Apple M4 Pro (20c GPU, 48GB)Apple Silicon Apple·FP16 | 6 | 3.2img/min | 48 GB | 50 W | $2.5k | — | — | Amazon | |
| #30 | NVIDIA Jetson AGX Orin 64GBEdge NVIDIA·FP16 | 6 | 1.8img/min | 64 GB | 60 W | $2.0k | — | — | Amazon |
Sorted by Images / min (high → low)Tip: $/1k tok is a 3-year amortization of MSRP at sustained throughput.
Cite this benchmark
Use this in your paper, blog post, or comparison table.
BibTeX
@misc{myaihardware_sdxl-1024_2026,
title = {MyAI Bench: SDXL 1024x1024},
author = {{MyAIHardware Contributors}},
year = {2026},
url = {https://www.myaihardware.com/benchmarks/workload/sdxl-1024},
note = {Version 1.3, accessed 2026-08-27}
}APA
MyAIHardware Contributors. (2026). MyAI Bench: SDXL 1024x1024. MyAIHardware. Retrieved 2026-08-27, from https://www.myaihardware.com/benchmarks/workload/sdxl-1024
MLA
MyAIHardware Contributors. "MyAI Bench: SDXL 1024x1024." MyAIHardware, 2026, https://www.myaihardware.com/benchmarks/workload/sdxl-1024. Accessed 2026-08-27.
Plain text
MyAI Bench, SDXL 1024x1024. MyAIHardware Contributors, 2026. Version 1.3. https://www.myaihardware.com/benchmarks/workload/sdxl-1024 (accessed 2026-08-27).