LiteLLM alternative.
Managed LiteLLM without the YAML and the self-hosted proxy.

# config.yaml
# model_list:
#   - model_name: gpt-5.4
#     litellm_params:
#       model: openai/gpt-5.4
#       api_key: os.environ/OPENAI_API_KEY

import openai

client = openai.OpenAI(
    base_url="http://your-litellm-proxy:4000",
    api_key="sk-litellm-...",
)

r = client.chat.completions.create(
    model="gpt-5.4",
    messages=[{"role": "user", "content": "Hello"}],
)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.aigateway.sh/v1",
    api_key="sk-aig-...",
)

r = client.chat.completions.create(
    model="openai/gpt-5.4",
    messages=[{"role": "user", "content": "Hello"}],
)

• Managed at the edge — no proxy to host, no VPC to attach, no YAML to maintain.
• 47ms p50 edge overhead globally — LiteLLM self-hosted often adds 150ms+.
• Unified billing, caching, evals, replay baked in — no bolted-on stack.
• Sub-account API, cost-attribution tags, bring-your-own-keys — all first-class.
• OpenAI-compatible; point the OpenAI SDK at our base URL and ship.

• LiteLLM has a richer config-as-code router for bespoke routing rules; we expose routing via aliases + evals, which covers most cases.