AUTOMATIC1111/stable-diffusion-webui · error · Exception

Sampler {config.name} is not supported for SDXL

Error message

Sampler {config.name} is not supported for SDXL

What it means

create_sampler() in sd_samplers.py refuses to build a sampler whose config carries options['no_sdxl']=True when the loaded model reports is_sdxl. Some sampler algorithms were never adapted to SDXL's UNet/attention shapes, so they are explicitly blacklisted for SDXL checkpoints instead of failing deep inside k-diffusion.

Source

Thrown at modules/sd_samplers.py:39

samplers_hidden = {}


def find_sampler_config(name):
    if name is not None:
        config = all_samplers_map.get(name, None)
    else:
        config = all_samplers[0]

    return config


def create_sampler(name, model):
    config = find_sampler_config(name)

    assert config is not None, f'bad sampler name: {name}'

    if model.is_sdxl and config.options.get("no_sdxl", False):
        raise Exception(f"Sampler {config.name} is not supported for SDXL")

    sampler = config.constructor(model)
    sampler.config = config

    return sampler


def set_samplers():
    global samplers, samplers_for_img2img, samplers_hidden

    samplers_hidden = set(shared.opts.hide_samplers)
    samplers = all_samplers
    samplers_for_img2img = all_samplers

    samplers_map.clear()
    for sampler in all_samplers:
        samplers_map[sampler.name.lower()] = sampler.name
        for alias in sampler.aliases:

View on GitHub (pinned to 82a973c043)

Solutions

1. Switch to an SDXL-supported sampler such as DPM++ 2M, DPM++ SDE, Euler, or Euler a
2. Update the WebUI — sampler/SDXL compatibility improves between releases and some samplers gained SDXL support
3. If the sampler is required, use a non-SDXL (SD1.5/SD2.x) checkpoint
4. For API scripts, derive the sampler list dynamically or catch this error and fall back to 'DPM++ 2M'

Example fix

# before (API payload pinned to an SD1.5-only sampler)
payload = {"sampler_name": "PLMS", ...}
# after
payload = {"sampler_name": "DPM++ 2M", ...}

from modules import sd_samplers
def is_sampler_ok_for_model(name, model):
    cfg = sd_samplers.find_sampler_config(name)
    return cfg is not None and not (getattr(model, 'is_sdxl', False) and cfg.options.get('no_sdxl', False))

try:
    sampler = sd_samplers.create_sampler(name, model)
except Exception as e:
    if 'not supported for SDXL' in str(e):
        sampler = sd_samplers.create_sampler('DPM++ 2M', model)
    else:
        raise

When it happens

Trigger: Loading an SDXL checkpoint and requesting a sampler marked no_sdxl in all_samplers (historically PLMS and a few k-diffusion variants) via the UI dropdown, --sampler launch flag, or the API sampler_name field.

Common situations: Scripts or API payloads hardcoding a sampler name that worked for SD1.5 (e.g. 'PLMS') after switching to an SDXL model; stale saved default sampler in settings after swapping checkpoints.

Related errors

• Sampler not found
• Invalid encoded image
• always on script {alwayson_script_name} not found
• Cannot have a selectable script in the always on scripts par
• Init image not found

AI-assisted analysis of AUTOMATIC1111/stable-diffusion-webui@82a973c043 (2026-08-14). Data as JSON: /api/errors/2f5744b16e30f08c. Report an issue: GitHub.