whatmodel.sh / image / Stable Diffusion 3 Medium
Stable Diffusion 3 Medium
↗ share
+ contribute
Stable Diffusion 3 Medium is an open-weights image model from Stability AI, released June 2024. Its published size is 2B parameters. Licence: stabilityai-community.
Model information sources
Release date ↗ · Weights licence ↗ · Parameter count ↗
2B = the SD3 Medium MMDiT as Stability states it; the CLIP-L/CLIP-G/T5-XXL text encoders and VAE are not included.
Score within the LMArena text-to-image comparison set only.
$ how it compares
The image models ranked either side of Stable Diffusion 3 Medium, out of 197 with a score, then more from Stability AI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 147 |  Skywork UniPic 1.5B | 40±6 | comparison only | ||
| 148 | Stable Diffusion XL 0.9 (base) | 40±6 | local | ||
| 149 | Stable Diffusion 3 Medium | 40±2 | api $0.035/image +local | ||
| 150 |  Kolors | 39±3 | local | ||
| 151 |  BLIP3-o 8B | 38±3 | local | ||
| more from Stability AI | |||||
| 137 | Stable Diffusion 3.5 Large | 42±1 | api $0.065/image +local | ||
| 141 | Stable Diffusion 3.5 Medium | 41±3 | api $0.035/image +local | ||
| 152 | Stable Cascade | 38±4 | local | ||
$ what it's good at
text to image▂▅▅▃▄▄█▃▁▂▁▁ #94 of 126 · 98% of its score · 4 tests
Beats Latent Diffusion text2img-large (LDM-KL-8-G, LAION-400M), VQ-Diffusion (ITHQ), Stable Diffusion XL 0.9 (base) here, though they rank higher overall. Trails SDXL-Lightning (4-step UNet), Stable Cascade here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| T2I-ReasonBench (overall) | 52.1% ✎ | ▅▂█▂▂▂····▂▂ beats 4/13 | 47% |
| EvalMuse-40K model ranking (FGA-BLIP2 overall alignment) | 3.27 / 5 ✎ | █▄·▄██▄▄█▄▄▄ beats 11/14 | 38% |
| WISE (WiScore overall, GPT-4o judge) | 42.0% ✎ | ▄·▆▂▆█▄▆▂▄·▄ beats 8/22 | 11% |
| HPSv3 image generation benchmark (Overall HPSv3 score) | 5.3100 ✎ | ▃·····▃··▃·█ beats 0/5 | 1.8% |
$ where to run it
Through an API
api · fal.ai · $0.035/image · fal.ai ↗ · price sourceUSD per generated image: fal bills fal-ai/stable-diffusion-v3-medium at $0.035 per image (billing unit images; fal-hosted serverless deployment of Stable Diffusion 3 Medium). Product: fal.ai serverless text-to-image. API model: fal-ai/stable-diffusion-v3-medium (Stable Diffusion 3 Medium; the page's image-to-image sibling hides its pricing and is not used). Settings: text-to-image, default size, one image. Price basis: published per-image list rate. Stability AI's own API no longer serves this checkpoint: its API deprecation notice of April 17, 2025 transitioned sd3-medium to sd3.5-medium. Currency: USD. Not a Stability AI direct tariff. fal.ai official model page, observed 2026-09-26.
Other ways to run it
version
- Run it as published · needs ~19.4 GB · from github.com/leejet
sd-cli -m models/sd3_medium_incl_clips_t5xxlfp16.safetensors -H 1024 -W 1024 -p 'a lovely cat holding a sign says "Stable Diffusion CPP"' --cfg-scale 4.5 --sampling-method euler -v --clip-on-cpu
Official weights and documentation: huggingface.co/stabilityai ↗
$ how the score was worked out · T2I-ReasonBench: Benchmarking… 47%, EvalMuse-40K: A Reliable and… 38%, WISE: A World Knowledge-Infor… 11% and 1 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 T2I-ReasonBench: Benchmarking Reasoning-Informed Text-to-Image Generation (Findings of ACL 2026) | 1 | 47% |
 EvalMuse-40K: A Reliable and Fine-Grained Benchmark with Comprehensive Human Annotations for Text-to-Image Generation Model Evaluation (arXiv v2) | 1 | 38% |
| WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation (arXiv v4) | 1 | 11% |
 HPSv3 repository README, image generation benchmark table (MizzenAI/HPSv3) | 1 | 2% |
| prior · pull toward the middle | 2% | |
| weighted average | = 39.5 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json