4 ms·Still 256K input tokens. So disappointing (predictable, but disappointing).by scottndecker 10mo agoStill 256K input tokens. So disappointing (predictable, but disappointing).htrp 10mo agomuch harder to train longer context inputscoder543 10mo agohttps://platform.openai.com/docs/models/gpt-5.2 https://platform.openai.com/docs/models/gpt-5.2 400k, not 256k.nathants 10mo ago400 - 128 = 272. Codex cli source.coder543 10mo agoIf you want to be able to generate up to 128k tokens in one go successfully, then yes, that math checks out.