BerriAI/litellm · error · ValueError

Error calling litellm.completion for generate_content: {e}

Error message

Error calling litellm.completion for generate_content: {e}

What it means

Catch-all from the sync generate_content adapter: every exception thrown by the underlying litellm.completion call (and the streaming-transform raise above it) is caught and re-raised as ValueError('Error calling litellm.completion for generate_content: <original>'). Diagnose the appended original message, not the wrapper.

Source

Thrown at litellm/google_genai/adapters/handler.py:163

                    )
                    return generate_content_response
                else:
                    # Transform streaming completion response to generate_content format
                    transformed_stream: Final = GOOGLE_GENAI_ADAPTER.translate_completion_output_params_streaming(
                        completion_response
                    )
                    if transformed_stream is not None:
                        return transformed_stream
                    raise ValueError("Failed to transform streaming response")
            else:
                # Transform completion response back to generate_content format
                generate_content_response = GOOGLE_GENAI_ADAPTER.translate_completion_to_generate_content(
                    cast(ModelResponse, completion_response)
                )
                return generate_content_response

        except Exception as e:
            raise ValueError(f"Error calling litellm.completion for generate_content: {e}")

View on GitHub (pinned to 6c2dcb801b)

Solutions

Inspect the suffix of the message / full traceback to find the true underlying error and fix that
Verify the model string and that the required API key env var is set for the resolved provider
Simplify: reproduce with litellm.completion(model='gemini/<model>', messages=[...]) to see the raw provider error
Update litellm if the failure is inside the transform itself

Example fix

# before
except ValueError as e:
    log(str(e))

# after
import traceback
except ValueError:
    traceback.print_exc()  # original exception is chained
    raise

Defensive patterns

Strategy: try-catch

Try / catch

try:
    r = generate_content(model=m, contents=c)
except ValueError as e:
    # suffix after 'generate_content: ' is the real provider error
    root = str(e).rsplit(": ", 1)[-1]
    handle(root)

Prevention

Set required GEMINI_API_KEY/GOOGLE_API_KEY in env
Reproduce failures with litellm.completion directly to bypass the wrapper

When it happens

Trigger: Sync generate_content calls that fail inside litellm.completion: bad/missing API keys, contents that cannot map to OpenAI messages, unknown model→provider mapping, or the streaming transform ValueError from the same try block.

Common situations: Switching from the official google-genai SDK to litellm's adapter and carrying over unsupported kwargs; API key env var naming mismatches (GEMINI_API_KEY vs GOOGLE_API_KEY); stale model names.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/bfa826b1ac50a8ae. Report an issue: GitHub.