4 ms·GPT-5.6-terra used 48.5% more context than Mimo-2.5-pro3 points by GodelNumbering 3mo agoGodelNumbering 3mo agoSharing this as a datapoint, not a benchmark