Skip to main content
Use LiteLLM when you already rely on its proxy, budgets, or local router, but want AI Stats model coverage, analytics, and server tools behind it.

​
Direct Python call

pip install litellm

import os
from litellm import completion

response = completion(
    model="openai/openai/gpt-5-nano",
    api_key=os.environ["AI_STATS_API_KEY"],
    api_base="https://api.phaseo.app/v1",
    messages=[{"role": "user", "content": "Reply with only: ok"}],
)

print(response.choices[0].message.content)
LiteLLM commonly uses the openai/ provider prefix for OpenAI-compatible endpoints. The AI Stats model id remains openai/gpt-5-nano, so the LiteLLM model string may look like openai/openai/gpt-5-nano.

​
Proxy config

Use a model alias if you do not want application code to include the LiteLLM provider prefix: 
model_list:
  - model_name: ai-stats-gpt-5-nano
    litellm_params:
      model: openai/openai/gpt-5-nano
      api_key: os.environ/AI_STATS_API_KEY
      api_base: https://api.phaseo.app/v1
Then call the LiteLLM proxy with model: "ai-stats-gpt-5-nano".

​
Notes

  • Prefer LiteLLM aliases for production so model naming stays readable.
  • If you use AI Stats server tools, test the exact LiteLLM route because some proxy layers normalize or drop unknown tool types.
  • For pricing, decide whether your system of record is LiteLLM, AI Stats, or both.
Last modified on June 11, 2026