Dev framework · configuration guide + live pricing · sign up for trial credit
LiteLLM is a unified multi-model calling layer with custom OpenAI-compatible endpoints — a good internal model gateway.
LiteLLM in one line: openai/-prefixed models + api_base in config.yaml. Recommended models and their market floor prices are in the table below; the full catalog is at https://321ai.cc/en/models/.
| Setting | Value | Notes |
|---|---|---|
model | openai/<platform model name> | add the openai/ prefix in config.yaml |
api_base | https://321ai.cc/v1 | |
api_key | API key from the console |
1. Add an entry under model_list in config.yaml with the openai/ prefix
2. Fill in api_base and api_key
3. Start the LiteLLM proxy and point clients at it
| What | Where | Notes |
|---|---|---|
| Live market floor pricing | https://321ai.cc/en/models/ | input per 1M tokens, per-supplier quotes on the marketplace |
Prices are the multi-supplier market floor and move in realtime; see all model prices →
It tells LiteLLM to speak the OpenAI-compatible protocol — it is not a platform requirement.