GTP-5 "The response ended prematurely"

[heltime]

When using gtp-5 on the request endpoint I consistently get an HttpIoException " The response ended prematurely. (ResponseEnded)" on non-trivial prompts after ca 60 seconds.

Setup:

• Timeouts are set to 600s
• I test on localhost, so no could infra in play
• Lib v2.4.0
• When streaming (OpenAIResponseClient. CreateResponseStreamingAsync), it happens consistently after ca 60s
• Non-streaming (OpenAIResponseClient. CreateResponseAsync) after 240s (tries and fails four times)
• Prompt length: 25302 characters (estimated 34645 tokens)

Other models work fine, only gtp-5 behaves like this. Has anyone else experienced this? Is this the API cutting the chord, or could there be an issue in the library? If I'm the only on seeing this, I guess error is on my end.