Learning Record learning from practice

· ai

OpenAI vs LiteLLM: API Cheat Sheet

#openai #litellm #python

The following table maps common OpenAI Python client calls to LiteLLM SDK calls.

The LiteLLM model name usually includes the provider prefix, for example openai/gpt-5.5.

API mapping

client.responses.create

OpenAI:

from openai import OpenAI

client = OpenAI()
response = client.responses.create(
    model="gpt-5.5",
    input="Say hello.",
)
print(response.output_text)

LiteLLM:

import litellm

response = litellm.responses(
    model="openai/gpt-5.5",
    input="Say hello.",
)
print(response)

client.models.list

OpenAI:

models = client.models.list()
print([model.id for model in models.data])

LiteLLM SDK has no universal equivalent because the available models belong to different providers. With LiteLLM Proxy, query its OpenAI-compatible /models endpoint using the OpenAI client:

proxy_client = OpenAI(
    api_key="your-proxy-key",
    base_url="http://localhost:4000/v1",
)
models = proxy_client.models.list()
print([model.id for model in models.data])

client.embeddings.create

OpenAI:

embedding_response = client.embeddings.create(
    model="text-embedding-3-small",
    input="Text to embed",
)
print(len(embedding_response.data[0].embedding))

LiteLLM:

embedding_response = litellm.embedding(
    model="openai/text-embedding-3-small",
    input=["Text to embed"],
)
print(len(embedding_response.data[0]["embedding"]))

client.chat.completions.create

OpenAI:

completion = client.chat.completions.create(
    model="gpt-5.5",
    messages=[
        {"role": "system", "content": "You are concise."},
        {"role": "user", "content": "Say hello."},
    ],
)
print(completion.choices[0].message.content)

LiteLLM:

response = litellm.completion(
    model="openai/gpt-5.5",
    messages=[
        {"role": "system", "content": "You are concise."},
        {"role": "user", "content": "Say hello."},
    ],
)
print(response.choices[0].message.content)

The request shape is similar, but provider support and provider-specific parameters can differ. Check the selected model and provider before assuming complete compatibility.

References