Struggling with Python's GIL and Multithreading for I/O-bound tasks – Am I missing something fundamental?

I'm trying to optimize an I/O-bound Python script that makes numerous external API calls concurrently. Despite using `concurrent.futures.ThreadPoolExecutor`, I'm not seeing the performance boost I expected, and it feels like the GIL is still a bottleneck even with I/O operations. Is there a common pitfall or a more effective pattern for genuinely parallelizing network requests in Python without moving to `asyncio` just yet?