Getting 400 INVALID_ARGUMENT Intermittently

Gemini API
, , ,
1

Error:

google.genai.errors.ClientError: 400 INVALID_ARGUMENT. {‘error’: {‘code’: 400, ‘message’: ‘Request contains an invalid argument.’, ‘status’: ‘INVALID_ARGUMENT’}}
at google/genai/errors.py:210 (raise_error_async)
at google/genai/errors.py:188 (raise_for_async_response)
at google/genai/_api_client.py:1312 (_async_request_once)
at tenacity/asyncio/init.py:116 (call)
at concurrent/futures/_base.py:401 (get_result)
at concurrent/futures/_base.py:449 (result)
at tenacity/init.py:184 (reraise)
at tenacity/init.py:413 (exc_check)
at tenacity/_utils.py:111 (inner)
at tenacity/asyncio/init.py:157 (iter)
at tenacity/asyncio/init.py:112 (call)
at google/genai/_api_client.py:1367 (_async_request)
at google/genai/_api_client.py:1434 (async_request)
at google/genai/models.py:5839 (_generate_content)
at google/genai/models.py:7033 (generate_content)
at app/services/gemini_service.py:264 (_call_gemini_api_with_config)
at app/utils/retry_utils.py:84 (retry_with_backoff)
at app/utils/retry_utils.py:89 (retry_with_backoff)
at app/utils/retry_utils.py:112 (wrapper)
at app/services/gemini_service.py:277 (_call_gemini_api_with_rotation)
at app/services/gemini_service.py:104 (generate_images_async)
at app/services/gemini_service.py:163 (generate_images_async)  # re-raise
at app/tracing/wrappers.py:53 (async_wrapper)
at app/worker.py:257 (_process_task_internal)
at app/worker.py:195 (process_task_async)
at app/worker.py:349 (run_job)

This was intermittent and failed for a while with the same inputs, then started working again without any changes on our side.

One thing I realized is that we’re sending MPO images but setting the MIME type to image/jpeg. That said, we’ve successfully generated thousands of images with Gemini using MPO inputs labeled as JPEG, so this hasn’t been an issue before.

I’m wondering if there’s anything we can do on our end to make this more reliable or avoid this kind of intermittent INVALID_ARGUMENT error going forward.

2

I am also facing the same issue. Need a fix asap we are losing trust with our customer guys

3

Flagged to the team, looking into this!

4

I am also starting to get this issue in the past two days.

5

We’re seeing a similar thing, in the nodejs geminiClient.files.upload() function. Intermittent 400 errors with INVALID_ARGUMENT. Retrying the request with the exact same data succeeds 95% of the time (remaining 5% need to be retried twice).

This is in a production app that’s been uploading thousands of files a month since November 25 without issue. These 400 errors started appearing about 24 hours ago for us.

1 Like