Getting absolutely jumbled responses back that make no sense. Using strcutured outputs and its returning things not even close to the specified schema. Anyone else having these issues?

You can use a fine-tuning model with either endpoint. You should be able to create identical token inputs (except in the case of currently broken ft:gpt-4.1 vision billing), and get similar output, verifying if one endpoint was symptomatic (like how gpt-3.5-turbo was somehow broken on only Chat Comp…

[image] _j: on any of t Well wouldnt there be justification b/c it wouldn’t mess up json schema?

A structured output for responses provided as a JSON schema can be supplied to either endpoint. Chat Completions: "response_format" Responses: "text.format" They also take slightly different shapes, with some of the container keys moved around. I’ve got some scripts to make API switch less of a h…

Oh I meant using the responses endpoint not chat completions. I’m pretty sure its recomendded not to do that for ft

For context I am using this with chat completions ‘response_format’: {‘type’: ‘json_schema’, ‘json_schema’: {‘name’: ‘angels_extraction’, ‘schema’: {‘type’: ‘object’, ‘properties’: {‘investors’: {‘type’: ‘array’, ‘items’: {‘type’: ‘object’, ‘properties’: {‘angel_investor_name’: {‘type’: ‘strin…

Important to note: JSON mode and JSON schema mode highly encourages “pretty” multi-line output. This was a poor decision by OpenAI, has no “off” switch in a “strict” enforcement, and the tabs and linefeeds trained in doing so is a direct cause of this persistent looping symptom, seen since gpt-4-t…

Thanks for the super in depth reply will look into it more. What I’m moreso worried about is this was working just fine for months and suddenly stopped working… I’ve trained on empty results as well. I’ve got a bunch of retries baked in if the json doesnt parse so it catches most of them. But where…

I’m honestly probably just going to implement an output parser that validates response format then retries if it fails. Which is kind stupid b/c thats what response_format is supposed to do.

A “strict”: true response format is logit enforcement, that should make it impossible to write anything but a valid JSON with the ordered keys, even if you train against it. However, it can still go haywire when in strings. OpenAI also doesn’t seem to enforce a closing token where it is needed, thu…

Degraded Performance on gpt-4.1-mini ft

API Bugs

toastywaffles888 December 10, 2025, 8:24pm 14

from playground too

Chat Completion responses suddenly returning malformed or inconsistent JSON

Topic		Replies	Views
Chat Completion responses suddenly returning malformed or inconsistent JSON Bugs bug , api , json	25	1137	December 15, 2025
Response has valid json but it's nested in broken json Bugs	16	4441	September 9, 2024
Function calling looping uncontrollably and calling unnecessarily Bugs function-calling , gpt-4o , gpt-4o-mini	27	4434	September 19, 2024
All Background Tasks on Responses API producing completely empty output array across all prompts Bugs api , responses-api , background-mode	6	686	September 17, 2025
Infinite wait for OpenAI server response with GPT5 on /completions under specific conditions API gpt5	13	1372	January 12, 2026

Degraded Performance on gpt-4.1-mini ft

Related topics