Run using the wrong assistant

I created two assistants in a thread.
When I create a run with the 2nd assistant I created:
“id”: “run_1234567”,
“object”: “”,
“created_at”: 1711972897,
“assistant_id”: “asst_1234567b”,
“thread_id”: “thread_1234567”,
“status”: “queued”,
“started_at”: null,
“expires_at”: 1711973497,
“cancelled_at”: null,
“failed_at”: null,
“completed_at”: null,
“required_action”: null,
“last_error”: null,
“model”: “gpt-3.5-turbo-0125”,
“instructions”: "Write a short summary of the text. All important key points must appear in the summary. The summary should not contain more than 100 words. ",
“tools”: ,
“file_ids”: ,
“metadata”: {},
“temperature”: 1.0,
“usage”: null
After the run has been completed the last message always uses the first assistant I created and not the one I passed when creating the run object. If I use the playground with the 2nd assistant, the result is Ok and what I expect using the 2nd assistant.

I this a know issue or is something wrong with my call to the API?

P.S.: It seems to me that if I pass a message a 2nd time to the thread, no matter what assistant ID I pass with a run, the result will always be that of the first assistant used on this message