Hi everyone, I'm Adrian. First off, thank you all for this community—I've learned so much from the discussions here.
I've been integrating the ChatGPT API into a small CI/CD notification tool I'm building in Python. Mostly, it works wonderfully. However, I've run into a persistent issue: when I send a particularly long prompt—like a detailed build log summary for analysis—the API sometimes returns a response that is completely empty. The HTTP status is 200, so it's not an error, but the `content` field in the response choices is just an empty string.
I'm using the standard `openai` Python library with `gpt-4-turbo`. My prompts are definitely under the token limit, and I'm handling streaming and non-streaming calls the same way. It seems to happen more often when the prompt includes structured data like JSON snippets or longer error traces.
Has anyone else encountered this? I'm wondering if it's a known timeout or a parsing issue on their end, or if there's something specific in the prompt formatting I should avoid. Any guidance would be greatly appreciated
still learning