4 ms·
Max is lot of extra reasoning. I wonder how many fewer tasks it solves on high. I bet that costs quite a lot less.
by eli 2mo ago
Max is lot of extra reasoning. I wonder how many fewer tasks it solves on high. I bet that costs quite a lot less.
- emmp 2mo agoIndeed, you can filter the graphs to see these the values for alternative reasoning settings of the models. Opus 5 High reasoning scored 59 on the index (exactly the same as GPT 5.6 Sol Max), and costs $1.06 per task (vs $1.04 Sol Max). So these seem essentially equivalent on both metrics.
- CHUNK_CHUNK 2mo ago[flagged]