diff --git a/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py b/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py index 6bddad09f21..ff56a5ec23c 100644 --- a/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py +++ b/litellm/llms/anthropic/experimental_pass_through/adapters/streaming_iterator.py @@ -129,8 +129,6 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): if should_start_new_block and not self.sent_content_block_finish: # Queue the sequence: content_block_stop -> content_block_start - # The trigger chunk itself is not emitted as a delta since the - # content_block_start already carries the relevant information. self.chunk_queue.append( { "type": "content_block_stop", @@ -145,6 +143,13 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): } ) self.sent_content_block_finish = False + # Also forward the triggering delta if it has content. + # When Bedrock omits contentBlockStart events (e.g. for marketplace + # models), the first delta for a new block arrives without a prior + # start event. Dropping it causes the first characters of each block + # to be silently lost. + if processed_chunk.get("type") == "content_block_delta": + self.chunk_queue.append(processed_chunk) return self.chunk_queue.popleft() if ( @@ -305,8 +310,6 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): if not self.queued_usage_chunk: if should_start_new_block and not self.sent_content_block_finish: # Queue the sequence: content_block_stop -> content_block_start - # The trigger chunk itself is not emitted as a delta since the - # content_block_start already carries the relevant information. # 1. Stop current content block self.chunk_queue.append( @@ -328,6 +331,14 @@ class AnthropicStreamWrapper(AdapterCompletionStreamWrapper): # Reset state for new block self.sent_content_block_finish = False + # Also forward the triggering delta if it has content. + # When Bedrock omits contentBlockStart events (e.g. for marketplace + # models), the first delta for a new block arrives without a prior + # start event. Dropping it causes the first characters of each block + # to be silently lost. + if processed_chunk.get("type") == "content_block_delta": + self.chunk_queue.append(processed_chunk) + # Return the first queued item return self.chunk_queue.popleft()