Assistant Responses problem

API
1

I am develop an application using the API of Assistants. Follwing the API documentation, I first create an assistant and then a thread. When user sends a message, a message call is send and then a run call with strem set to true.

My problem is in events thread.run.created, thread.run.queued, thread.run.in_progress, thread.message.completed, the response contains the instructions to assistant, which I do not need them and they are large. How can I avoide to receive them in every single message? Below are my API Calls and the responses. Thanks in advance.

Assistant:

URL: APIURL/assistants

Body: {“model”:“gpt-4o”,“name”:“Oddy bot”,“instructions”:“Open the document on /link to google drive document/ and follow the instructions written on it”}

JAVA Response: {“id”:“asst_TjDxUkmPG3oVomxIvyRLmkPB”,“object”:“assistant”,“created_at”:1720162239,“name”:“Oddy bot”,“description”:null,“model”:“gpt-4o”,“instructions”:“Open the document on /link to google drive document/ and follow the instructions written on it”,“tools”:,“top_p”:1,“temperature”:1,“tool_resources”:{},“metadata”:{},“response_format”:“auto”}

Thread:

URL:APIURL/threads

Body:

JAVA Response: {“id”:“thread_GpCYOuVUO9JtFVuyCHiZoIyd”,“object”:“thread”,“created_at”:1720162240,“metadata”:{},“tool_resources”:{}}

Message:

URL:APIURL/threads/thread_GpCYOuVUO9JtFVuyCHiZoIyd/messages

Body: {“role”:“user”,“content”:“Καλημέρα”}

JAVA Response: {“id”:“msg_G54kYosxLYUkxt3t80d3UAjO”,“object”:“thread.message”,“created_at”:1720162300,“assistant_id”:null,“thread_id”:“thread_GpCYOuVUO9JtFVuyCHiZoIyd”,“run_id”:null,“role”:“user”,“content”:[{“type”:“text”,“text”:{“value”:“Καλημέρα”,“annotations”:}}],“attachments”:,“metadata”:{}}

Run:

URL:APIURL/threads/thread_GpCYOuVUO9JtFVuyCHiZoIyd/runs

Body: {“assistant_id”:“asst_TjDxUkmPG3oVomxIvyRLmkPB”,“stream”:true}

Raw Response Line: event: thread.run.created

Raw Response Line: data: {“id”:“run_qDfcHJn3MmX1JNbtrnyszWhq”,“object”:“thread.run”,“created_at”:1720162300,“assistant_id”:“asst_TjDxUkmPG3oVomxIvyRLmkPB”,“thread_id”:“thread_GpCYOuVUO9JtFVuyCHiZoIyd”,“status”:“queued”,“started_at”:null,“expires_at”:1720162900,“cancelled_at”:null,“failed_at”:null,“completed_at”:null,“required_action”:null,“last_error”:null,“model”:“gpt-4o”,“instructions”:“Open the document on /link to google drive document/ and follow the instructions written on it”,“tools”:,“tool_resources”:{“code_interpreter”:{“file_ids”:}},“metadata”:{},“temperature”:1.0,“top_p”:1.0,“max_completion_tokens”:null,“max_prompt_tokens”:null,“truncation_strategy”:{“type”:“auto”,“last_messages”:null},“incomplete_details”:null,“usage”:null,“response_format”:“auto”,“tool_choice”:“auto”,“parallel_tool_calls”:true}

Raw Response Line: event: thread.run.queued

Raw Response Line: data: {“id”:“run_qDfcHJn3MmX1JNbtrnyszWhq”,“object”:“thread.run”,“created_at”:1720162300,“assistant_id”:“asst_TjDxUkmPG3oVomxIvyRLmkPB”,“thread_id”:“thread_GpCYOuVUO9JtFVuyCHiZoIyd”,“status”:“queued”,“started_at”:null,“expires_at”:1720162900,“cancelled_at”:null,“failed_at”:null,“completed_at”:null,“required_action”:null,“last_error”:null,“model”:“gpt-4o”,“instructions”:“Open the document on /link to google drive document/ and follow the instructions written on it”,“tools”:,“tool_resources”:{“code_interpreter”:{“file_ids”:}},“metadata”:{},“temperature”:1.0,“top_p”:1.0,“max_completion_tokens”:null,“max_prompt_tokens”:null,“truncation_strategy”:{“type”:“auto”,“last_messages”:null},“incomplete_details”:null,“usage”:null,“response_format”:“auto”,“tool_choice”:“auto”,“parallel_tool_calls”:true}

…

2

It seems there is no way to specify which values are part of the stream object.
The API reference has “null” for instructions, but obviously there are part of the format so that they can be used at some point in the streaming of steps.
Since it comes at steps before the AI response, there may be no slowdown or delay, because the AI also has to initialize before it starts producing tokens.