LiteLLM provides a common interface for model calls. Its built-in mock_response can return a fixed answer while you check the calling code. This example verifies the response text and chat.completion shape without contacting a provider.
What the isolated test checked
The model name selects an interface configuration; the named model was not called. First check response access in the application, then verify authentication, errors and retries with the chosen provider. This success does not establish proxy routing or fallback behavior.
The following is the recorded synthetic output, not a model benchmark.
{
"model": "gpt-4o-mini",
"content": "08:30",
"object": "chat.completion"
}
Reproduce the small task
The recorded runtime was Linux with Python 3.12. Installation requires public package downloads; execution was offline. Windows shell behavior was not tested. Use a disposable environment for the example.
python -m pip install litellm==1.109.1
import sys,json,traceback,importlib.metadata as meta,pathlib,asyncio,subprocess,os
SLUG='litellm'
def emit(package,checks,observations,**kw):
print(json.dumps({'checks':checks,'observations':observations},ensure_ascii=False,default=str,indent=2))
def litellm():
import litellm
r=litellm.completion(model='openai/gpt-4o-mini',messages=[{'role':'user','content':'When does NOTE-31 leave?'}],mock_response='08:30')
emit('litellm',dict(content=r.choices[0].message.content=='08:30',shape=r.object=='chat.completion'),dict(model=r.model,content=r.choices[0].message.content,object=r.object),scope='Built-in mock_response normalizes a synthetic completion; no live provider, proxy routing, failover, latency or prices validated.')
litellm()
Limits and untested behavior
Providers, proxy operation, fallbacks, prices, rate limits and latency remain untested. A mock produces no real inference bill and cannot validate production cost.
Other capabilities described in the pinned official documentation below were not locally tested. Source availability alone does not establish free hosted execution or free model calls.
Sources and test record
Codex ran the synthetic task in a non-root, network-disabled container without publishing credentials or private inputs. This is not a claim that the site owner personally tested the tool.
Tested package: 1.109.1. Documentation commit: 8c97c243feb8. The registry package and README snapshot are separately pinned, not assumed identical. Test date: 2026-11-15.