I’m building a small agent around Grok’s API and want it to fetch two independent pieces of information at once. I can handle multiple tool results in my application, but I can’t tell whether the model can request several tools in the same turn or whether I should expect one call, wait for the result, and then ask again.

If you’ve implemented parallel function calls with Grok, how did you handle the returned tool-call IDs and send the results back? I’m curious whether there are any practical differences from the usual OpenAI-compatible tool-calling flow.