BerriAI/litellm · error · ValueError

'model' keyword not found and unable to extract model from e

Error message

'model' keyword not found and unable to extract model from endpoint. Expected format: /model/{modelId}/{action}. Got: {endpoint}

What it means

Bedrock model-ID extraction helper failed: the endpoint path contains neither a literal 'model' path segment nor any known BEDROCK_ENDPOINT_ACTIONS segment from which the model ID could be inferred, so the path does not match the expected /model/{modelId}/{action} shape. Thrown as ValueError during URL parsing before any request is sent.

Source

Thrown at litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py:680

                break

        # If "model" keyword not found, try to extract model from the endpoint
        # by finding the action and taking everything before it
        if model_index is None:
            # Find the index of the action in the endpoint parts
            action_index = None
            for idx, part in enumerate(endpoint_parts):
                if part in BEDROCK_ENDPOINT_ACTIONS:
                    action_index = idx
                    break

            if action_index is not None and action_index > 1:
                # Join all parts before the action (excluding empty strings)
                model_parts = [p for p in endpoint_parts[1:action_index] if p]
                if model_parts:
                    return "/".join(model_parts)

            raise ValueError(
                f"'model' keyword not found and unable to extract model from endpoint. Expected format: /model/{{modelId}}/{{action}}. Got: {endpoint}"
            )

        # Find the index of the action in the endpoint parts
        action_index = None
        for idx, part in enumerate(endpoint_parts):
            if part in BEDROCK_ENDPOINT_ACTIONS:
                action_index = idx
                break

        if action_index is not None and action_index > model_index + 1:
            # Join all parts between "model" and the action (excluding "model" itself)
            return "/".join(endpoint_parts[model_index + 1 : action_index])

        # Fallback to taking everything after "model" if no action found
        model_parts = [p for p in endpoint_parts[model_index + 1 :] if p]
        if model_parts:
            return "/".join(model_parts)

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Format the Bedrock pass-through endpoint as /model/{modelId}/{action}, e.g. /model/anthropic.claude-v2/invoke.
2. URL-encode the model ID if it contains slashes (use %2F).

When it happens

Trigger: Thrown at litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py:680 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/1ab1070cff334686. Report an issue: GitHub.