BerriAI/litellm · error · ValueError

Missing Novita AI API Key - A call is being made to novita b

Error message

Missing Novita AI API Key - A call is being made to novita but no key is set either in the environment variables or via params

What it means

Raised by NovitaConfig.validate_environment when api_key is None: a call to a novita/* model was made without an api_key parameter. Unlike most providers, this check does not fall back to an environment variable lookup inside the transformer — litellm's caller layer normally resolves the NOVITA_API_KEY env var, so hitting this means the key was not supplied anywhere.

Source

Thrown at litellm/llms/novita/chat/transformation.py:25

"""

from ....types.llms.openai import AllMessageValues
from ...openai.chat.gpt_transformation import OpenAIGPTConfig


class NovitaConfig(OpenAIGPTConfig):
    def validate_environment(
        self,
        headers: dict,
        model: str,
        messages: list[AllMessageValues],
        optional_params: dict,
        litellm_params: dict,
        api_key: str | None = None,
        api_base: str | None = None,
    ) -> dict:
        if api_key is None:
            raise ValueError(
                "Missing Novita AI API Key - A call is being made to novita but no key is set either in the environment variables or via params"
            )
        headers["Authorization"] = f"Bearer {api_key}"
        headers["Content-Type"] = "application/json"
        headers["X-Novita-Source"] = "litellm"
        return headers

View on GitHub (pinned to 6c2dcb801b)

Solutions

export NOVITA_API_KEY=<key from novita.ai> and re-run.
Or pass api_key explicitly to the completion call.
In litellm proxy, set api_key under the novita model's litellm_params in config.yaml.
Print bool(os.getenv('NOVITA_API_KEY')) to confirm the env var is actually visible to the process.

Example fix

# before
out = litellm.completion(model="novita/meta-llama/Llama-3-8B-Instruct", messages=msgs)

# after
out = litellm.completion(
    model="novita/meta-llama/Llama-3-8B-Instruct",
    messages=msgs,
    api_key=os.environ["NOVITA_API_KEY"],
)

Defensive patterns

Strategy: validation

Validate before calling

import os
if not os.getenv("NOVITA_API_KEY"):
    raise RuntimeError("NOVITA_API_KEY not set (get one at novita.ai)")

Try / catch

try:
    out = litellm.completion(model="novita/meta-llama/Llama-3-8B-Instruct", messages=msgs)
except ValueError as e:
    if "Missing Novita AI API Key" in str(e):
        raise RuntimeError("Config error: NOVITA_API_KEY missing") from e
    raise

Prevention

Assert provider env vars at startup for every model you route to.
In proxy configs, set api_key per novita deployment explicitly.
Automate secret injection (k8s secrets, SSM) instead of manual exports.

When it happens

Trigger: Calling litellm.completion(model='novita/...') with no api_key argument and no NOVITA_API_KEY environment variable set (litellm would otherwise inject it as the api_key param).

Common situations: Env var missing in the deployed environment (container, serverless, CI), typo in NOVITA_API_KEY, or proxy config missing the key in the model's litellm_params.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/bb4a88282e556696. Report an issue: GitHub.