Gpt-4o seems to be broken for assistants after the rollout of gpt-4o-mini

After the recent update to GPT 4o-mini, it seems like the assistants are broken on gpt-4o while they work fine on GPT-4o-mini. The same assistants were working properly on 4o/4 earlier and now they do not seem to follow hte instructions where as 4o-mini is now working as 4o earlier. But it seems like 4o-mini just forgets the context half way through the conversation.