BerriAI/litellm · error · ValueError

Chat provider: Invalid text delta {parsed_chunk}

Error message

Chat provider: Invalid text delta {parsed_chunk}

What it means

For response.output_text.delta events the transformer requires a non-None 'delta' field with the text fragment; if the event lacks it, it raises. Symmetric to the function-arguments case: the provider streamed a text-delta event without any text payload.

Source

Thrown at litellm/completion_extras/litellm_responses_transformation/transformation.py:1381

                        )
                    ]
                )

        elif event_type == "response.output_text.delta":
            # Content part added to output
            content_part = parsed_chunk.get("delta", None)
            if content_part is not None:
                return ModelResponseStream(
                    choices=[
                        StreamingChoices(
                            index=0,
                            delta=Delta(content=content_part),
                            finish_reason=None,
                        )
                    ]
                )
            else:
                raise ValueError(f"Chat provider: Invalid text delta {parsed_chunk}")
        elif event_type == "response.reasoning_summary_text.delta":
            content_part = parsed_chunk.get("delta", None)
            if content_part:
                return ModelResponseStream(
                    choices=[
                        StreamingChoices(
                            index=cast(int, parsed_chunk.get("summary_index")),
                            delta=Delta(reasoning_content=content_part),
                        )
                    ]
                )
        elif event_type == "response.completed":
            # Response is fully complete - now we can signal is_finished=True
            # This ensures we don't prematurely end the stream before tool_calls arrive

            # Check if response contains function_call items in output
            # to determine correct finish_reason
            response_data: Final = parsed_chunk.get("response", {})

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Drop text-delta events whose delta is None before transformation
2. Treat delta == '' as valid (it is) — only None/missing trips the error
3. Update or report the gateway producing the malformed event

Example fix

# before
for ev in events:
    chunks = transformer.chunk_parser(ev, ...)

# after
for ev in events:
    if ev.get("type") == "response.output_text.delta" and ev.get("delta") is None:
        continue
    chunks = transformer.chunk_parser(ev, ...)

def text_delta_event_valid(ev: dict) -> bool:
    return ev.get("type") != "response.output_text.delta" or ev.get("delta") is not None

try:
    out = transformer.chunk_parser(ev, ...)
except ValueError as e:
    if "Invalid text delta" in str(e):
        out = None  # tolerate empty text-delta events
    else:
        raise

When it happens

Trigger: A response.output_text.delta event arrives with delta missing or None — non-conformant providers, proxies rewriting events, or malformed fixtures.

Common situations: Gateway emulation quirks; recorded SSE fixtures missing fields; providers that emit empty text deltas as null instead of empty strings.

Related errors

• Chat provider: Invalid function argument delta {parsed_chunk
• Chat provider: Empty parsed_chunk
• Chat provider: Invalid chunk type {type(parsed_chunk)}
• Braintrust API error: {e.response.text}
• Failed to connect to Braintrust API: {str(e)}

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/874fa668484d5a97. Report an issue: GitHub.