[Bug]: openai.BadRequestError: We could not parse the JSON body of your request

[dobin]

Do you need to file an issue?

• I have searched the existing issues and this bug is not already filed.
• My model is hosted on OpenAI or Azure. If not, please look at the "model providers" issue and don't file a new one here.
• I believe this is a legitimate bug, not just a question. If this is a question, please use the Discussions area.

Describe the bug

When indexing, i get the following error message:

2026-03-21 06:03:32.0180 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71525/106718
2026-03-21 06:03:32.0180 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71526/106718
2026-03-21 06:03:32.0180 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71527/106718
2026-03-21 06:03:32.0180 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71528/106718
2026-03-21 06:03:32.0180 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71529/106718
2026-03-21 06:03:32.0180 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71530/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71531/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71532/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71533/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71534/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71535/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71536/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71537/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71538/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71539/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71540/106718
2026-03-21 06:03:32.0181 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71541/106718
2026-03-21 06:03:32.0182 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71542/106718
2026-03-21 06:03:32.0303 - ERROR - graphrag_llm.middleware.with_logging - Async request failed with exception=litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 929, in acompletion
    headers, response = await self.make_openai_chat_completion_request(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/logging_utils.py", line 297, in async_wrapper
    result = await func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 460, in make_openai_chat_completion_request
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 437, in make_openai_chat_completion_request
    await openai_aclient.chat.completions.with_raw_response.create(
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_legacy_response.py", line 384, in wrapped
    return cast(LegacyAPIResponse[R], await func(*args, **kwargs))
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/resources/chat/completions/completions.py", line 2714, in create
    return await self._post(
           ^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1884, in post
    return await self.request(cast_to, opts, stream=stream, stream_cls=stream_cls)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1669, in request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 620, in acompletion
    response = await init_response
               ^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 989, in acompletion
    raise OpenAIError(
litellm.llms.openai.common_utils.OpenAIError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_logging.py", line 64, in _request_count_middleware_async
    return await async_middleware(**kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_request_count.py", line 72, in _request_count_middleware_async
    result = await async_middleware(**kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_cache.py", line 145, in _cache_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_retries.py", line 55, in _retry_middleware_async
    return await retrier.retry_async(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/retry/exponential_retry.py", line 102, in retry_async
    return await func(**input_args)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_metrics.py", line 80, in _metrics_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/completion/lite_llm_completion.py", line 290, in _base_completion_async
    response = await litellm.acompletion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 2072, in wrapper_async
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 1892, in wrapper_async
    result = await original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 639, in acompletion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2456, in exception_type
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 478, in exception_type
    raise BadRequestError(
litellm.exceptions.BadRequestError: litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
2026-03-21 06:03:32.0324 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71543/106718
2026-03-21 06:03:32.0325 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71544/106718
2026-03-21 06:03:32.0715 - ERROR - graphrag.index.run.run_pipeline - error running workflow extract_graph
Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 929, in acompletion
    headers, response = await self.make_openai_chat_completion_request(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/logging_utils.py", line 297, in async_wrapper
    result = await func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 460, in make_openai_chat_completion_request
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 437, in make_openai_chat_completion_request
    await openai_aclient.chat.completions.with_raw_response.create(
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_legacy_response.py", line 384, in wrapped
    return cast(LegacyAPIResponse[R], await func(*args, **kwargs))
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/resources/chat/completions/completions.py", line 2714, in create
    return await self._post(
           ^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1884, in post
    return await self.request(cast_to, opts, stream=stream, stream_cls=stream_cls)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1669, in request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 620, in acompletion
    response = await init_response
               ^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 989, in acompletion
    raise OpenAIError(
litellm.llms.openai.common_utils.OpenAIError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/run/run_pipeline.py", line 135, in _run_pipeline
    result = await workflow_function(config, context)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/extract_graph.py", line 61, in run_workflow
    entities, relationships, raw_entities, raw_relationships = await extract_graph(
                                                               ^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/extract_graph.py", line 143, in extract_graph
    entities, relationships = await get_summarized_entities_relationships(
                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/extract_graph.py", line 168, in get_summarized_entities_relationships
    entity_summaries, relationship_summaries = await summarize_descriptions(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 115, in summarize_descriptions
    return await get_summarized(entities_df, relationships_df, semaphore)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 80, in get_summarized
    edge_results = await asyncio.gather(*edge_futures)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/asyncio/tasks.py", line 385, in __wakeup
    future.result()
  File "/home/dobin/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/asyncio/tasks.py", line 314, in __step_run_and_handle_result
    result = coro.send(None)
             ^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 102, in do_summarize_descriptions
    results = await run_summarize_descriptions(
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 139, in run_summarize_descriptions
    result = await extractor(id=id, descriptions=descriptions)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/description_summary_extractor.py", line 68, in __call__
    result = await self._summarize_descriptions(id, descriptions)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/description_summary_extractor.py", line 105, in _summarize_descriptions
    result = await self._summarize_descriptions_with_llm(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/description_summary_extractor.py", line 124, in _summarize_descriptions_with_llm
    response: LLMCompletionResponse = await self._model.completion_async(
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/completion/lite_llm_completion.py", line 193, in completion_async
    response = await self._completion_async(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_logging.py", line 64, in _request_count_middleware_async
    return await async_middleware(**kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_request_count.py", line 72, in _request_count_middleware_async
    result = await async_middleware(**kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_cache.py", line 145, in _cache_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_retries.py", line 55, in _retry_middleware_async
    return await retrier.retry_async(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/retry/exponential_retry.py", line 102, in retry_async
    return await func(**input_args)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_metrics.py", line 80, in _metrics_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/completion/lite_llm_completion.py", line 290, in _base_completion_async
    response = await litellm.acompletion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 2072, in wrapper_async
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 1892, in wrapper_async
    result = await original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 639, in acompletion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2456, in exception_type
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 478, in exception_type
    raise BadRequestError(
litellm.exceptions.BadRequestError: litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
2026-03-21 06:03:32.0737 - ERROR - graphrag.api.index - Workflow extract_graph completed with errors
2026-03-21 06:03:32.0740 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 71545/106718
2026-03-21 06:03:32.0744 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/text-embedding-3-small: {
  "attempted_request_count": 1,
  "successful_response_count": 1,
  "failed_response_count": 0,
  "failure_rate": 0.0,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 2.2788126468658447,
  "compute_duration_per_response_seconds": 2.2788126468658447,
  "cache_hit_rate": 0.0,
  "streaming_responses": 0,
  "responses_with_tokens": 1,
  "prompt_tokens": 9,
  "total_tokens": 9,
  "tokens_per_response": 9.0,
  "cost_per_response": 0.0
}
2026-03-21 06:03:32.0745 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/gpt-5.4-mini: {
  "attempted_request_count": 21831,
  "successful_response_count": 21829,
  "failed_response_count": 2,
  "failure_rate": 9.161284412074573e-05,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 48129.54413795471,
  "compute_duration_per_response_seconds": 2.204844204404907,
  "cached_responses": 6,
  "cache_hit_rate": 0.0002748637134087682,
  "streaming_responses": 0,
  "responses_with_tokens": 21829,
  "prompt_tokens": 22409866,
  "completion_tokens": 8838469,
  "total_tokens": 31248335,
  "tokens_per_response": 1431.5055659901966,
  "cost_per_response": 0.0
}

Steps to reproduce

graphrag init
git clone https://github.com/dobin/AwesomeMalDevLinks
mkdir input/
mv AwesomeMalDevLinks/out/maldev/*.md input/maldev
graphrag index

Expected Behavior

No error

GraphRAG Config Used

I changed model, and file pattern:

completion_models:
  default_completion_model:
    model_provider: openai
    model: gpt-5.4-mini
    auth_method: api_key # or azure_managed_identity
    api_key: ${GRAPHRAG_API_KEY} # set this in the generated .env file, or remove if managed identity
    retry:
      type: exponential_backoff

embedding_models:
  default_embedding_model:
    model_provider: openai
    model: text-embedding-3-small
    auth_method: api_key
    api_key: ${GRAPHRAG_API_KEY}
    retry:
      type: exponential_backoff

input:
  type: text # [csv, text, json, jsonl]
  file_pattern: ".*\\.md$$"

Logs and screenshots

No response

Additional Information

• GraphRAG Version: 3.0.6
• Operating System: ubuntu wsl 22.04
• Python Version: 3.12.13
• Related Issues:

[dobin]

Using only 100 of the 400 documents, i get:

(bookmarkRag) dobin@paddy:~/repos/bookmarkRag$ graphrag index
Starting pipeline with workflows: load_input_documents, create_base_text_units, create_final_documents, extract_graph, finalize_graph, extract_covariates, create_communities, create_final_text_units, create_community_reports, generate_text_embeddings
Starting workflow: load_input_documents

Workflow complete: load_input_documents
Starting workflow: create_base_text_units
  100 / 100 ........................................................................................
Workflow complete: create_base_text_units
Starting workflow: create_final_documents

Workflow complete: create_final_documents
Starting workflow: extract_graph
/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/extract_graph/extract_graph.py:119: FutureWarning: The behavior of DataFrame concatenation with empty or all-NA entries is deprecated. In a future version, this will no longer exclude empty or all-NA columns when determining the result dtypes. To retain the old behavior, exclude the relevant entries before the concat operation.
  all_relationships = pd.concat(relationship_dfs, ignore_index=False)
  31689 / 31689 ....................................................................................
Workflow complete: extract_graph
Starting workflow: finalize_graph

Workflow complete: finalize_graph
Starting workflow: extract_covariates

Workflow complete: extract_covariates
Starting workflow: create_communities

Workflow complete: create_communities
Starting workflow: create_final_text_units

Workflow complete: create_final_text_units
Starting workflow: create_community_reports
  2357 / 2357 ......................................................................................
Workflow complete: create_community_reports
Starting workflow: generate_text_embeddings
[2026-03-21T06:33:45Z WARN  lance::dataset::write::insert] No existing dataset at /home/dobin/repos/bookmarkRag/output/lancedb/entity_description.lance, it will be created
Pipeline error: Column 1 named vector expected length 400 but got length 200........................
Pipeline complete
(bookmarkRag) dobin@paddy:~/repos/bookmarkRag$ 

2026-03-21 07:33:54.0200 - INFO - graphrag.logger.progress - generate embeddings progress: 24/25
2026-03-21 07:33:54.0201 - INFO - graphrag.logger.progress - generate embeddings progress: 25/25
2026-03-21 07:33:54.0664 - ERROR - graphrag.index.run.run_pipeline - error running workflow generate_text_embeddings
Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/run/run_pipeline.py", line 135, in _run_pipeline
    result = await workflow_function(config, context)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/generate_text_embeddings.py", line 89, in run_workflow
    await generate_text_embeddings(
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/generate_text_embeddings.py", line 145, in generate_text_embeddings
    count = await embed_text(
            ^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/embed_text/embed_text.py", line 60, in embed_text
    total_rows += await _flush_embedding_buffer(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/embed_text/embed_text.py", line 136, in _flush_embedding_buffer
    vector_store.load_documents(documents)
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_vectors/lancedb.py", line 109, in load_documents
    data = pa.table({
           ^^^^^^^^^^
  File "pyarrow/table.pxi", line 6210, in pyarrow.lib.table
  File "pyarrow/table.pxi", line 1985, in pyarrow.lib._Tabular.from_pydict
  File "pyarrow/table.pxi", line 6402, in pyarrow.lib._from_pydict
  File "pyarrow/table.pxi", line 4912, in pyarrow.lib.Table.from_arrays
  File "pyarrow/table.pxi", line 4256, in pyarrow.lib.Table.validate
  File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Column 1 named vector expected length 400 but got length 200
2026-03-21 07:33:54.0679 - ERROR - graphrag.api.index - Workflow generate_text_embeddings completed with errors
2026-03-21 07:33:54.0682 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/text-embedding-3-small: {
  "attempted_request_count": 26,
  "successful_response_count": 26,
  "failed_response_count": 0,
  "failure_rate": 0.0,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 91.4250659942627,
  "compute_duration_per_response_seconds": 3.5163486920870266,
  "cache_hit_rate": 0.0,
  "streaming_responses": 0,
  "responses_with_tokens": 26,
  "prompt_tokens": 43941,
  "total_tokens": 43941,
  "tokens_per_response": 1690.0384615384614,
  "cost_per_response": 0.0
}
2026-03-21 07:33:54.0682 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/gpt-5.4-mini: {
  "attempted_request_count": 9612,
  "successful_response_count": 9611,
  "failed_response_count": 1,
  "failure_rate": 0.00010403662089055348,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 31115.348031044006,
  "compute_duration_per_response_seconds": 3.237472482680679,
  "cached_responses": 6,
  "cache_hit_rate": 0.0006242846738112579,
  "streaming_responses": 0,
  "responses_with_tokens": 9611,
  "prompt_tokens": 13661375,
  "completion_tokens": 5315704,
  "total_tokens": 18977079,
  "tokens_per_response": 1974.5165955675789,
  "cost_per_response": 0.0
}

I am confused - does this work at all?

[dobin]

I set to large embedding model, which allowed the initial index with a few documents:

    model: text-embedding-3-large

Adding more results in:

(bookmarkRag) dobin@paddy:~/repos/bookmarkRag$ graphrag update
Starting pipeline with workflows: load_update_documents, create_base_text_units, create_final_documents, extract_graph, finalize_graph, extract_covariates, create_communities, create_final_text_units, create_community_reports, generate_text_embeddings, update_final_documents, update_entities_relationships, update_text_units, update_covariates, update_communities, update_community_reports, update_text_embeddings, update_clean_state
Starting workflow: load_update_documents

Workflow complete: load_update_documents
Starting workflow: create_base_text_units
  100 / 100 ........................................................................................
Workflow complete: create_base_text_units
Starting workflow: create_final_documents

Workflow complete: create_final_documents
Starting workflow: extract_graph
/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/extract_graph/extract_graph.py:119: FutureWarning: The behavior of DataFrame concatenation with empty or all-NA entries is deprecated. In a future version, this will no longer exclude empty or all-NA columns when determining the result dtypes. To retain the old behavior, exclude the relevant entries before the concat operation.
  all_relationships = pd.concat(relationship_dfs, ignore_index=False)
Pipeline error: litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
Pipeline complete

[maocatooo]

Recently, I have also been assigned to investigate this issue
graphrag==1.2.0
LLMmodel:gpt-4o-mini

It looks like a problem with OPENAI

[dobin]

I upgraded the packages and tried again. I got a different error, which i dont know if its related to the other errors.

But it seems to work overall.

(bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ ls -1 input/ | wc -l
127

$ graphrag index
Starting pipeline with workflows: load_input_documents, create_base_text_units, create_final_documents, extract_graph, finalize_graph, extract_covariates, create_communities, create_final_text_units, create_community_reports, generate_text_embeddings
Starting workflow: load_input_documents

Workflow complete: load_input_documents
Starting workflow: create_base_text_units
  127 / 127 ........................................................................................
Workflow complete: create_base_text_units
Starting workflow: create_final_documents

Workflow complete: create_final_documents
Starting workflow: extract_graph
/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/extract_graph/extract_graph.py:119: FutureWarning: The behavior of DataFrame concatenation with empty or all-NA entries is deprecated. In a future version, this will no longer exclude empty or all-NA columns when determining the result dtypes. To retain the old behavior, exclude the relevant entries before the concat operation.
  all_relationships = pd.concat(relationship_dfs, ignore_index=False)
  33899 / 33899 ....................................................................................
Workflow complete: extract_graph
Starting workflow: finalize_graph

Workflow complete: finalize_graph
Starting workflow: extract_covariates

Workflow complete: extract_covariates
Starting workflow: create_communities

Workflow complete: create_communities
Starting workflow: create_final_text_units

Workflow complete: create_final_text_units
Starting workflow: create_community_reports
  389 / 2414 .......................................................................................
  2414 / 2414 ......................................................................................
Workflow complete: create_community_reports
Starting workflow: generate_text_embeddings
[2026-04-02T04:43:04Z WARN  lance::dataset::write::insert] No existing dataset at /home/dobin/repos/bookmarkRag/grag/maldev/output/lancedb/entity_description.lance, it will be created
[2026-04-02T04:46:45Z WARN  lance::dataset::write::insert] No existing dataset at /home/dobin/repos/bookmarkRag/grag/maldev/output/lancedb/community_full_content.lance, it will be created
Unclosed client session.............................................................................
client_session: <aiohttp.client.ClientSession object at 0x745164d6d580>
Unclosed connector
connections: ['deque([(<aiohttp.client_proto.ResponseHandler object at 0x74515db62930>, 32709.800434037), (<aiohttp.client_proto.ResponseHandler object at 0x74515db62450>, 32709.988942524), (<aiohttp.client_proto.ResponseHandler object at 0x74515db62630>, 32709.989709251), (<aiohttp.client_proto.ResponseHandler object at 0x74515db63470>, 32709.989823997), (<aiohttp.client_proto.ResponseHandler object at 0x74515db62c90>, 32709.989952404), (<aiohttp.client_proto.ResponseHandler object at 0x74515db63050>, 32710.187924604), (<aiohttp.client_proto.ResponseHandler object at 0x74515db61490>, 32710.388499346), (<aiohttp.client_proto.ResponseHandler object at 0x74515db63cb0>, 32711.132192276), (<aiohttp.client_proto.ResponseHandler object at 0x74515db60cb0>, 32711.328280797), (<aiohttp.client_proto.ResponseHandler object at 0x74515dac8470>, 32711.515711044), (<aiohttp.client_proto.ResponseHandler object at 0x74515db63dd0>, 32711.704323099), (<aiohttp.client_proto.ResponseHandler object at 0x74515db60d70>, 32711.894660487), (<aiohttp.client_proto.ResponseHandler object at 0x74515db61b50>, 32712.089979748), (<aiohttp.client_proto.ResponseHandler object at 0x74515db60dd0>, 32712.284388509), (<aiohttp.client_proto.ResponseHandler object at 0x74515db616d0>, 32712.470089156), (<aiohttp.client_proto.ResponseHandler object at 0x74515db60a70>, 32712.653563625), (<aiohttp.client_proto.ResponseHandler object at 0x74515db623f0>, 32712.836201428), (<aiohttp.client_proto.ResponseHandler object at 0x74515db621b0>, 32713.027047865), (<aiohttp.client_proto.ResponseHandler object at 0x74515db62810>, 32713.210546971), (<aiohttp.client_proto.ResponseHandler object at 0x74515db63230>, 32713.398742828), (<aiohttp.client_proto.ResponseHandler object at 0x74515db62e10>, 32713.588034146), (<aiohttp.client_proto.ResponseHandler object at 0x74515db62570>, 32713.776961003), (<aiohttp.client_proto.ResponseHandler object at 0x74515db63110>, 32713.963731304), (<aiohttp.client_proto.ResponseHandler object at 0x74515db60e90>, 32714.151479381), (<aiohttp.client_proto.ResponseHandler object at 0x74515dac8c50>, 32714.345643947)])']
connector: <aiohttp.connector.TCPConnector object at 0x745164d6f3b0>
[2026-04-02T04:50:09Z WARN  lance::dataset::write::insert] No existing dataset at /home/dobin/repos/bookmarkRag/grag/maldev/output/lancedb/text_unit_text.lance, it will be created
  8 / 8 ............................................................................................
Workflow complete: generate_text_embeddings
Pipeline complete

bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ uv pip show graphrag openai litellm
Using Python 3.12.13 environment at: /home/dobin/repos/bookmarkRag/.venv
Name: graphrag
Version: 3.0.8
Location: /home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages
Requires: azure-identity, azure-search-documents, azure-storage-blob, blis, devtools, graphrag-cache, graphrag-chunking, graphrag-common, graphrag-input, graphrag-llm, graphrag-storage, graphrag-vectors, graspologic-native, json-repair, networkx, nltk, numpy, pandas, pyarrow, pydantic, spacy, textblob, tqdm, typer, typing-extensions
Required-by:
---
Name: litellm
Version: 1.82.6
Location: /home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages
Requires: aiohttp, click, fastuuid, httpx, importlib-metadata, jinja2, jsonschema, openai, pydantic, python-dotenv, tiktoken, tokenizers
Required-by: graphrag-llm
---
Name: openai
Version: 2.30.0
Location: /home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages
Requires: anyio, distro, httpx, jiter, pydantic, sniffio, tqdm, typing-extensions
Required-by: litellm

...

2026-04-02 06:51:01.0527 - INFO - graphrag.index.operations.embed_text.run_embed_text - embedding 57 inputs via 57 snippets using 8 batches. max_batch_size=16, batch_max_tokens=8191
2026-04-02 06:51:04.0612 - INFO - graphrag.logger.progress - generate embeddings progress: 1/8
2026-04-02 06:51:04.0877 - INFO - graphrag.logger.progress - generate embeddings progress: 2/8
2026-04-02 06:51:04.0882 - INFO - graphrag.logger.progress - generate embeddings progress: 3/8
2026-04-02 06:51:04.0889 - INFO - graphrag.logger.progress - generate embeddings progress: 4/8
2026-04-02 06:51:04.0895 - INFO - graphrag.logger.progress - generate embeddings progress: 5/8
2026-04-02 06:51:04.0898 - INFO - graphrag.logger.progress - generate embeddings progress: 6/8
2026-04-02 06:51:04.0900 - INFO - graphrag.logger.progress - generate embeddings progress: 7/8
2026-04-02 06:51:04.0906 - INFO - graphrag.logger.progress - generate embeddings progress: 8/8
2026-04-02 06:51:04.0998 - INFO - graphrag.index.workflows.generate_text_embeddings - Embedded 857 rows for text_unit_text
2026-04-02 06:51:05.0000 - INFO - graphrag.index.workflows.generate_text_embeddings - Workflow completed: generate_text_embeddings
2026-04-02 06:51:05.0016 - INFO - graphrag.api.index - Workflow generate_text_embeddings completed successfully
2026-04-02 06:51:05.0027 - INFO - graphrag.index.run.run_pipeline - Indexing pipeline complete.
2026-04-02 06:51:05.0060 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/text-embedding-3-small: {
  "attempted_request_count": 1261,
  "successful_response_count": 1261,
  "failed_response_count": 0,
  "failure_rate": 0.0,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 3610.259944677353,
  "compute_duration_per_response_seconds": 2.863013437491953,
  "cache_hit_rate": 0.0,
  "streaming_responses": 0,
  "responses_with_tokens": 1261,
  "prompt_tokens": 4190509,
  "total_tokens": 4190509,
  "tokens_per_response": 3323.163362410785,
  "cost_per_response": 0.0
}
2026-04-02 06:51:05.0061 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/gpt-5.4-mini: {
  "attempted_request_count": 10047,
  "successful_response_count": 10047,
  "failed_response_count": 0,
  "failure_rate": 0.0,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 42145.471967458725,
  "compute_duration_per_response_seconds": 4.194831488748753,
  "cached_responses": 6,
  "cache_hit_rate": 0.0005971931919976113,
  "streaming_responses": 0,
  "responses_with_tokens": 10047,
  "prompt_tokens": 14664967,
  "completion_tokens": 5609181,
  "total_tokens": 20274148,
  "tokens_per_response": 2017.9305265253308,
  "cost_per_response": 0.0
}

[dobin]

Using, yet again, 300 documents, it fails yet again. This is getting expensive.

(bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ mv ../../input_2/* input/
(bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ rmdir ../../input_2/
(bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ graphrag update
Starting pipeline with workflows: load_update_documents, create_base_text_units, create_final_documents, extract_graph, finalize_graph, extract_covariates, create_communities, create_final_text_units, create_community_reports, generate_text_embeddings, update_final_documents, update_entities_relationships, update_text_units, update_covariates, update_communities, update_community_reports, update_text_embeddings, update_clean_state
Starting workflow: load_update_documents

Workflow complete: load_update_documents
Starting workflow: create_base_text_units
  213 / 213 ........................................................................................
Workflow complete: create_base_text_units
Starting workflow: create_final_documents

Workflow complete: create_final_documents
Starting workflow: extract_graph
/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/extract_graph/extract_graph.py:119: FutureWarning: The behavior of DataFrame concatenation with empty or all-NA entries is deprecated. In a future version, this will no longer exclude empty or all-NA columns when determining the result dtypes. To retain the old behavior, exclude the relevant entries before the concat operation.
  all_relationships = pd.concat(relationship_dfs, ignore_index=False)
Pipeline error: litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
Pipeline complete
(bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ ls^C
(bookmarkRag) dobin@paddy:~/repos/bookmarkRag/grag/maldev$ ls -l input/ | wc -l
441

2026-04-02 08:46:48.0152 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1088/53531
2026-04-02 08:46:48.0191 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1089/53531
2026-04-02 08:46:48.0289 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1090/53531
2026-04-02 08:46:48.0290 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1091/53531
2026-04-02 08:46:48.0349 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1092/53531
2026-04-02 08:46:48.0416 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1093/53531
2026-04-02 08:46:48.0450 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1094/53531
2026-04-02 08:46:48.0457 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1095/53531
2026-04-02 08:46:48.0493 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1096/53531
2026-04-02 08:46:48.0596 - ERROR - graphrag_llm.middleware.with_logging - Async request failed with exception=litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 929, in acompletion
    headers, response = await self.make_openai_chat_completion_request(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/logging_utils.py", line 297, in async_wrapper
    result = await func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 460, in make_openai_chat_completion_request
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 437, in make_openai_chat_completion_request
    await openai_aclient.chat.completions.with_raw_response.create(
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_legacy_response.py", line 384, in wrapped
    return cast(LegacyAPIResponse[R], await func(*args, **kwargs))
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/resources/chat/completions/completions.py", line 2714, in create
    return await self._post(
           ^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1884, in post
    return await self.request(cast_to, opts, stream=stream, stream_cls=stream_cls)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1669, in request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 620, in acompletion
    response = await init_response
               ^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 989, in acompletion
    raise OpenAIError(
litellm.llms.openai.common_utils.OpenAIError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_logging.py", line 64, in _request_count_middleware_async
    return await async_middleware(**kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_request_count.py", line 72, in _request_count_middleware_async
    result = await async_middleware(**kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_cache.py", line 145, in _cache_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_retries.py", line 55, in _retry_middleware_async
    return await retrier.retry_async(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/retry/exponential_retry.py", line 102, in retry_async
    return await func(**input_args)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_metrics.py", line 80, in _metrics_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/completion/lite_llm_completion.py", line 290, in _base_completion_async
    response = await litellm.acompletion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 2094, in wrapper_async
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 1893, in wrapper_async
    result = await original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 639, in acompletion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2456, in exception_type
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 478, in exception_type
    raise BadRequestError(
litellm.exceptions.BadRequestError: litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
2026-04-02 08:46:48.0818 - ERROR - graphrag.index.run.run_pipeline - error running workflow extract_graph
Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 929, in acompletion
    headers, response = await self.make_openai_chat_completion_request(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/logging_utils.py", line 297, in async_wrapper
    result = await func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 460, in make_openai_chat_completion_request
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 437, in make_openai_chat_completion_request
    await openai_aclient.chat.completions.with_raw_response.create(
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_legacy_response.py", line 384, in wrapped
    return cast(LegacyAPIResponse[R], await func(*args, **kwargs))
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/resources/chat/completions/completions.py", line 2714, in create
    return await self._post(
           ^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1884, in post
    return await self.request(cast_to, opts, stream=stream, stream_cls=stream_cls)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/openai/_base_client.py", line 1669, in request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 620, in acompletion
    response = await init_response
               ^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/llms/openai/openai.py", line 989, in acompletion
    raise OpenAIError(
litellm.llms.openai.common_utils.OpenAIError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/run/run_pipeline.py", line 135, in _run_pipeline
    result = await workflow_function(config, context)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/extract_graph.py", line 61, in run_workflow
    entities, relationships, raw_entities, raw_relationships = await extract_graph(
                                                               ^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/extract_graph.py", line 143, in extract_graph
    entities, relationships = await get_summarized_entities_relationships(
                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/workflows/extract_graph.py", line 168, in get_summarized_entities_relationships
    entity_summaries, relationship_summaries = await summarize_descriptions(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 115, in summarize_descriptions
    return await get_summarized(entities_df, relationships_df, semaphore)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 60, in get_summarized
    node_results = await asyncio.gather(*node_futures)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/asyncio/tasks.py", line 385, in __wakeup
    future.result()
  File "/home/dobin/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/asyncio/tasks.py", line 314, in __step_run_and_handle_result
    result = coro.send(None)
             ^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 102, in do_summarize_descriptions
    results = await run_summarize_descriptions(
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/summarize_descriptions.py", line 139, in run_summarize_descriptions
    result = await extractor(id=id, descriptions=descriptions)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/description_summary_extractor.py", line 68, in __call__
    result = await self._summarize_descriptions(id, descriptions)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/description_summary_extractor.py", line 105, in _summarize_descriptions
    result = await self._summarize_descriptions_with_llm(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag/index/operations/summarize_descriptions/description_summary_extractor.py", line 124, in _summarize_descriptions_with_llm
    response: LLMCompletionResponse = await self._model.completion_async(
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/completion/lite_llm_completion.py", line 193, in completion_async
    response = await self._completion_async(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_logging.py", line 64, in _request_count_middleware_async
    return await async_middleware(**kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_request_count.py", line 72, in _request_count_middleware_async
    result = await async_middleware(**kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_cache.py", line 145, in _cache_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_retries.py", line 55, in _retry_middleware_async
    return await retrier.retry_async(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/retry/exponential_retry.py", line 102, in retry_async
    return await func(**input_args)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/middleware/with_metrics.py", line 80, in _metrics_middleware_async
    response = await async_middleware(**kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/graphrag_llm/completion/lite_llm_completion.py", line 290, in _base_completion_async
    response = await litellm.acompletion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 2094, in wrapper_async
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/utils.py", line 1893, in wrapper_async
    result = await original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/main.py", line 639, in acompletion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2456, in exception_type
    raise e
  File "/home/dobin/repos/bookmarkRag/.venv/lib/python3.12/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 478, in exception_type
    raise BadRequestError(
litellm.exceptions.BadRequestError: litellm.BadRequestError: OpenAIException - We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)
2026-04-02 08:46:48.0838 - ERROR - graphrag.api.index - Workflow extract_graph completed with errors
2026-04-02 08:46:48.0839 - INFO - graphrag.logger.progress - Summarize entity/relationship description progress: 1097/53531
2026-04-02 08:46:48.0845 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/text-embedding-3-small: {
  "attempted_request_count": 1,
  "successful_response_count": 1,
  "failed_response_count": 0,
  "failure_rate": 0.0,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 0.3553915023803711,
  "compute_duration_per_response_seconds": 0.3553915023803711,
  "cache_hit_rate": 0.0,
  "streaming_responses": 0,
  "responses_with_tokens": 1,
  "prompt_tokens": 9,
  "total_tokens": 9,
  "tokens_per_response": 9.0,
  "cost_per_response": 0.0
}
2026-04-02 08:46:48.0845 - INFO - graphrag_llm.metrics.log_metrics_writer - Metrics for openai/gpt-5.4-mini: {
  "attempted_request_count": 3291,
  "successful_response_count": 3290,
  "failed_response_count": 1,
  "failure_rate": 0.00030385900941962927,
  "requests_with_retries": 0,
  "retries": 0,
  "retry_rate": 0.0,
  "compute_duration_seconds": 20209.67073225975,
  "compute_duration_per_response_seconds": 6.142757061477128,
  "cache_hit_rate": 0.0,
  "streaming_responses": 0,
  "responses_with_tokens": 3290,
  "prompt_tokens": 9300024,
  "completion_tokens": 3492759,
  "total_tokens": 12792783,
  "tokens_per_response": 3888.3838905775074,
  "cost_per_response": 0.0
}

[maocatooo]

OH,I returned here from my email. I’ve now fully investigated and resolved this issue, and I hope it can be helpful to you.

The error We could not parse the JSON body of your request is thrown by the OpenAI SDK. In principle, requests made by the OpenAI SDK should not produce this error, but it still occurred. I believe this may be a bug in OpenAI.

My solution is to include this HTTP 400 error in the retry mechanism to handle the issue.

I’ve written an article about this (in Chinese):
https://github.com/maocatooo/maocatooo.github.io/blob/master/blog/ai/rag/graphrag_openai_json_err.md

My investigation was conducted on graphrag==1.2.0, so there may be some differences compared to your version.

[dworthen]

Hi @dobin

There seems to be two issues reported here.

1. Embedding sizes. If you use an embedding model that returns vector sizes that differ from the default of 3072 then you need to update the vector_store section of your config to define the expected vector_size. Specifically, you need to define a vector_size for text_unit_text, entity_description, and community_full_content. More details can be viewed in the docs: https://microsoft.github.io/graphrag/config/yaml/
2. JSON parsing. We just merged PR Support client side json validation. #2306 to hopefully improve JSON parsing for models that either don't support structured responses or are inconsistent with structured responses. This update will be available in our next pypi release.