Skip to content
← Models
Open weightsStability AI/10 trustMetrics-checked
Stable Diffusion XL logo

Stable Diffusion XL

The open-weights image workhorse: 4.3M downloads, giant fine-tune ecosystem.

Visit official page →

About

Stable Diffusion XL 1.0 (stabilityai/stable-diffusion-xl-base-1.0), released July 2023, is Stability AI's 2.6B-parameter text-to-image model — and the most-deployed open image model ever, with 4.3M Hugging Face downloads and the largest fine-tune ecosystem in existence (thousands of LoRAs, checkpoints, ControlNets, and tools like ComfyUI built around it). It runs on consumer GPUs and remains the default for self-hosted image generation. Licensed under CreativeML Open RAIL++-M — a use-restricted community license, not OSI open source; describe it as "open weights". Superseded on raw quality by SD3.5 and Flux, but unbeaten on ecosystem.

Specs

Context window

N/A (image model; 77-token CLIP text encoder prompts, extendable)

Pricing

Self-host: free (OpenRAIL++-M). Hosted inference via Replicate/fal.ai and others — per-image pricing varies; check live provider pages.

Free quota

No official API; self-hosting on your own GPU is free.

API access

Self-host (Diffusers, ComfyUI, Automatic1111, llama.cpp-compatible servers) or hosted inference APIs.

Weights

Open weights

Released

July 25, 2023

Benchmarks

No standard academic benchmarks; evaluated on photorealism/prompt adherence vs contemporaries (2023-2024). Its moat is ecosystem scale, not benchmark leadership — now dated vs Flux/SD3.5 on quality.

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

image generationfine-tuningLoRAcontrolnet

Trust score — 7/10

/10 trustMetrics-checked

Scored 7/10 (metrics tier, metrics-v1). Evidence: HF: stabilityai/stable-diffusion-xl-base-1.0; HF downloads 4,300,160 (peer-normalized); released 1176 days ago. Manual review: most-deployed open image model (4.3M HF dl, giant LoRA/ControlNet ecosystem, consumer-GPU friendly); superseded on quality by SD3.5/Flux. License: CreativeML Open RAIL++-M — use-restricted community license, NOT OSI open source; describe as 'open weights'.

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
9.6/10
Maintenance · recency, releases, issue load
4.0/10
Openness · license + docs
6.5/10
Manual review · sanity read + red-flag check
8.0/10
  • · HF: stabilityai/stable-diffusion-xl-base-1.0
  • · HF downloads 4,300,160 (peer-normalized)
  • · released 1176 days ago
  • · manual review: most-deployed open image model (4.3M HF dl, giant LoRA/ControlNet ecosystem, consumer-GPU friendly); superseded on quality by SD3.5/Flux

More in the sdxl family

Sources