BerriAI/litellm · error · Exception

Unable to map model response to known provider format. model

Error message

Unable to map model response to known provider format. model={model}

What it means

After dispatching the raw Bedrock responses to the per-provider _transform_response handlers, the dispatcher found returned_response still None — meaning the provider selected for the model produced no mapped EmbeddingResponse. This is an internal mapping gap (or an unrecognized response shape) rather than a user input error in the usual sense.

Source

Thrown at litellm/llms/bedrock/embed/embedding.py:223

                returned_response = AmazonTitanG1Config()._transform_response(response_list=response_list, model=model)
            elif model == "amazon.titan-embed-text-v2:0":
                returned_response = AmazonTitanV2Config()._transform_response(response_list=response_list, model=model)
            elif model == "amazon.titan-embed-g1-text-02":
                returned_response = AmazonTitanG1Config()._transform_response(response_list=response_list, model=model)
            elif provider == "twelvelabs":
                returned_response = TwelveLabsMarengoEmbeddingConfig()._transform_response(
                    response_list=response_list, model=model
                )
            elif provider == "nova":
                returned_response = AmazonNovaEmbeddingConfig()._transform_response(
                    response_list=response_list, model=model, batch_data=batch_data
                )

        ##########################################################
        # Validate returned response
        ##########################################################
        if returned_response is None:
            raise Exception(f"Unable to map model response to known provider format. model={model}")
        return returned_response

    def _single_func_embeddings(
        self,
        client: HTTPHandler | None,
        timeout: float | httpx.Timeout | None,
        batch_data: list[dict],
        credentials: Any,
        extra_headers: dict | None,
        endpoint_url: str,
        aws_region_name: str,
        model: str,
        logging_obj: Any,
        provider: BEDROCK_EMBEDDING_PROVIDERS_LITERAL,
        api_key: str | None = None,
        is_async_invoke: bool | None = False,
    ):
        responses: Final[list[dict]] = []

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Upgrade litellm to the latest release — new Bedrock embedding models are added frequently.
2. Print the exact model string passed (after async_invoke/ stripping) and compare with litellm's supported bedrock embedding providers (cohere, amazon, titan, twelvelabs, nova).
3. If the model is genuinely unsupported, switch to a supported equivalent (e.g. amazon.titan-embed-text-v2:0).
4. Report the model + raw response to the litellm repo if the model is listed as supported.

Example fix

# before
resp = litellm.embedding(model="bedrock/<brand-new-embedding-model>", input=["hi"])

# after
pip install -U litellm
resp = litellm.embedding(model="bedrock/amazon.titan-embed-text-v2:0", input=["hi"])

import litellm
from litellm.llms.bedrock.embed.embedding import BedrockEmbeddingConfig
SUPPORTED = {"cohere", "amazon", "titan", "twelvelabs", "nova"}
assert model.split("/")[-1].split(".")[0] in SUPPORTED or "titan" in model

def is_supported_bedrock_embedding_model(model: str) -> bool:
    stem = model.removeprefix("bedrock/").removeprefix("async_invoke/")
    return any(p in stem for p in ("cohere", "titan", "amazon", "twelvelabs", "nova"))

try:
    resp = litellm.embedding(model=model, input=["hi"])
except Exception as e:
    if "Unable to map model response" in str(e):
        log.warning("unsupported bedrock embedding model %s on litellm %s", model, litellm.__version__)
    raise

When it happens

Trigger: A model string that routes to a provider branch which returns None (e.g. a provider case that does not match, or a response payload shape the transformer does not recognize), so the final 'Validate returned response' guard fires.

Common situations: Using a newly released Bedrock embedding model not yet in the installed LiteLLM version's transformation map; a provider branch silently not setting returned_response; upgrading AWS model versions while running an older litellm release.

Related errors

• Invalid data URL format: {data_url[:50]}...
• Invalid data URL format (missing comma): {data_url[:50]}...
• output_s3_uri is required for async invoke requests
• No embedding data found in response: {response}
• Missing boto3 to call bedrock. Run 'pip install boto3'.

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/414758d6169bf154. Report an issue: GitHub.