BerriAI/litellm · error · HTTPException

Default Model not on proxy. Configure via `/model/new` or co

Error message

Default Model not on proxy. Configure via `/model/new` or config.yaml. Default_model={data.default_model}, proxy_model_names={set(llm_router.get_model_names())}

What it means

POST /customer/new rejects a default_model that is not among the router's model names (model_name entries in config.yaml model_list or models added via /model/new). The 422 detail echoes the requested default_model and the full proxy_model_names set, so the exact alias to use is visible in the error itself.

Source

Thrown at litellm/proxy/management_endpoints/customer_endpoints.py:419

        llm_router,
        prisma_client,
    )

    if prisma_client is None:
        raise HTTPException(
            status_code=500,
            detail={"error": CommonProxyErrors.db_not_connected_error.value},
        )
    try:
        ## VALIDATION ##
        if data.default_model is not None:
            if llm_router is None:
                raise HTTPException(
                    status_code=422,
                    detail={"error": CommonProxyErrors.no_llm_router.value},
                )
            elif data.default_model not in llm_router.get_model_names():
                raise HTTPException(
                    status_code=422,
                    detail={
                        "error": f"Default Model not on proxy. Configure via `/model/new` or config.yaml. Default_model={data.default_model}, proxy_model_names={set(llm_router.get_model_names())}"
                    },
                )

        new_end_user_obj: dict[str, object] = {}

        ## CREATE BUDGET ## if set
        _new_budget: Final = new_budget_request(data)
        if _new_budget is not None:
            try:
                budget_record: Final = await _typed_table(BudgetRepository(prisma_client)).create(
                    data={
                        **_new_budget.model_dump(exclude_unset=True),
                        "created_by": user_api_key_dict.user_id or litellm_proxy_admin_name,
                        "updated_by": user_api_key_dict.user_id or litellm_proxy_admin_name,
                    }

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Use an alias from the error's proxy_model_names set exactly as printed
2. Or add the model under the expected name: POST /model/new with model_name equal to the desired default_model
3. After creation, /customer/update can repoint default_model to an existing alias

Example fix

# before
curl -X POST http://localhost:4000/customer/new -d '{"user_id": "u1", "default_model": "gpt-4o"}'   # proxy only knows "openai/gpt-4o" -> 422

# after
curl -X POST http://localhost:4000/customer/new -d '{"user_id": "u1", "default_model": "openai/gpt-4o"}'

import httpx


def model_exists(base: str, headers: dict, model: str) -> bool:
    data = httpx.get(f"{base}/v1/models", headers=headers).json()["data"]
    return model in {m["id"] for m in data}


assert model_exists(BASE, HDR, payload["default_model"]) or "default_model" not in payload

try:
    r = httpx.post(f"{base}/customer/new", json=payload, headers=headers)
except httpx.HTTPStatusError as e:
    if e.response.status_code == 422 and "Default Model not on proxy" in e.response.text:
        # the detail lists valid proxy_model_names - pick one and retry
        raise ValueError(f"default_model invalid; pick from /v1/models") from e
    raise

When it happens

Trigger: Passing default_model "gpt-4o" when the proxy exposes it as "openai/gpt-4o" or an Azure deployment alias; typos; referencing a model that was removed from config while customer templates still name it.

Common situations: Teams migrating from the raw OpenAI SDK where provider ids worked directly; alias naming drift between environments (dev names openai/gpt-4o, prod names gpt4o-prod); retired deployments still referenced by onboarding scripts.

Related errors

• No models configured on proxy
• user_id is required, passed user_id = {data.user_id}
• user_id is required, passed user_id = {data.user_ids}
• max_budget cannot be negative. Received: {data.max_budget}
• soft_budget cannot be negative. Received: {data.soft_budget}

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/802c6ee2405e7beb. Report an issue: GitHub.