Structured Outputs with Batch Processing

API
4

@slippy I tried your example above using Batch API and it worked fine. It might be super silly question, but in your .jsonl you don’t have any indentation right?

Anyway, here are the exact steps I took :blush:

Step 1: I created a batchinput.jsonl with the following contents (NOTE: single line, no indentations, whitespaces are OK by JSON standard)

{"custom_id": "request-1", "method": "POST", "url": "/v1/chat/completions", "body": {"model": "gpt-4o-mini", "messages": [{"role": "system", "content": "You are a helpful math tutor. Guide the user through the solution step by step."}, {"role": "user", "content": "how can I solve 8x + 7 = -23"}], "max_tokens": 4096, "response_format": {"type": "json_schema", "json_schema": {"name": "math_response", "schema": {"type": "object", "properties": {"steps": {"type": "array", "items": {"type": "object", "properties": {"explanation": {"type": "string"}, "output": {"type": "string"}}, "required": ["explanation", "output"], "additionalProperties": false}}, "final_answer": {"type": "string"}}, "required": ["steps", "final_answer"], "additionalProperties": false}, "strict": true}}}}

Step 2: I uploaded batchinput.jsonl via Files API. I received a file ID in the response; in this case it is file-F2ieFWin68wvubNmBPOUvsDW.

curl https://api.openai.com/v1/files \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -F purpose="batch" \
  -F file="@batchinput.jsonl"

Step 3: I created a batch using the above file ID. I received a batch ID in the response; in this case it was batch_asVAZzeehZ4mf2QE9zi1krvE

curl https://api.openai.com/v1/batches \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input_file_id": "file-F2ieFWin68wvubNmBPOUvsDW",
    "endpoint": "/v1/chat/completions",
    "completion_window": "24h"
  }'

Step 4: I queried the batch status using the above batch ID. When I saw that the status was set to completed, I noted the output_file_id, in this case it is file-8E8AlSiy5puk3vBkthQ8UNlw.

curl https://api.openai.com/v1/batches/batch_asVAZzeehZ4mf2QE9zi1krvE \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json"

Step 5: Retrieve the final output using the above output file ID.

curl https://api.openai.com/v1/files/file-8E8AlSiy5puk3vBkthQ8UNlw/content \
  -H "Authorization: Bearer $OPENAI_API_KEY" > batch_output.jsonl

The final output (as per batch_output.jsonl) is as follows:

{"id": "batch_req_3aNHqQt8UB1idP6RkXetGJgM", "custom_id": "request-1", "response": {"status_code": 200, "request_id": "a7ba2b98e25ff47189d7550eee2d8072", "body": {"id": "chatcmpl-9xc57s5a7ngga92riZU0wpcIdhEG3", "object": "chat.completion", "created": 1723994765, "model": "gpt-4o-mini-2024-07-18", "choices": [{"index": 0, "message": {"role": "assistant", "content": "{\"steps\":[{\"explanation\":\"First, we will isolate the term with 'x' by moving the constant term to the other side of the equation. We can do this by subtracting 7 from both sides.\",\"output\":\"8x + 7 - 7 = -23 - 7\"},{\"explanation\":\"This simplifies to 8x = -30.\",\"output\":\"8x = -30\"},{\"explanation\":\"Next, we will isolate 'x' by dividing both sides of the equation by 8.\",\"output\":\"x = -30 / 8\"},{\"explanation\":\"Now we can simplify -30 / 8. We can divide both the numerator and the denominator by 2.\",\"output\":\"x = -15 / 4\"},{\"explanation\":\"Finally, we can write -15 / 4 in decimal form if necessary. -15 / 4 = -3.75.\",\"output\":\"x = -3.75\"}],\"final_answer\":\"x = -15/4 or x = -3.75\"}", "refusal": null}, "logprobs": null, "finish_reason": "stop"}], "usage": {"prompt_tokens": 45, "completion_tokens": 207, "total_tokens": 252}, "system_fingerprint": "fp_507c9469a1"}}, "error": null}
8 Likes