The mini variant in the new GPT-5 series of models from OpenAI.
Priced close to Gemini 2.5 Flash
#GPT-5-mini
pine jetty
upbeat river
is minimal a required param?
use case: i want to turn off reasoning, cuz for my use case, i dont need it
light lotus
Is this model's endpoint working? I swapped a call that used gpt-4.1-mini to gpt-5-mini and I get HTTP 400 errors. I'm using the same parameters I did with gpt-4.1-mini.
Maybe I'm just too early? 🙂
snow drift
so far I'm getting "worse than Gemini flash" answers
(which flash version? yeah, sometimes worse than 2.0)
pine jetty
Benching above Gemini 2.5 Flash in my own benchmark, SVGBench. Will post results later with other GPT-5 models
kind heath
anyone know how to pass encrypted reasoning in multi-turn conversation!!
granite raven
Benching above Gemini 2.5 Flash in my own benchmark, SVGBench. Will post results...
gpt-5 on grok-4 level seems about right. whats very strange about the result?
past oasis
gpt-5 on grok-4 level seems about right. whats very strange about the result?
i think he meant compared to the results of Horizon Alpha and Beta
pine jetty
gpt-5 on grok-4 level seems about right. whats very strange about the result?
GPT-5-mini is performing much better than GPT-5 on SVGBench, which goes against the usual "bigger model better" intuition
sonic tiger
Is this model good for coding like Horizon Alpha?