rc1
11 hours ago
Quite hidden in the mobile app at least. I had to click thinking. The slider goes 5.6 Instant, 5.6 Medium, 5.6 High, 5.6 Extra High, and the surprise 6 Pro
minimaxir
11 hours ago
I see this as well. Odd there is only 6 Pro.
11 hours ago
Quite hidden in the mobile app at least. I had to click thinking. The slider goes 5.6 Instant, 5.6 Medium, 5.6 High, 5.6 Extra High, and the surprise 6 Pro
11 hours ago
I see this as well. Odd there is only 6 Pro.
12 hours ago
Also is rolling out to the $100/$200 Codex subscriptions, I got access this morning.
Unsurprisingly, it does indeed consume usage at ~2.5x the rate of Sol.
11 hours ago
Shouldn't this new reduction in thinking output make the model cheaper to operate? Could it be that the model is the same old LLM, with a bit newer architecture but still doing the same things, including huge amounts of thinking, just that its thinking is less clear to human readers?
10 hours ago
Those tend to be under so-called ‘max’ modes, or ‘ultra’ - where you scale up inference time compute and then choose amongst your answers. Astra is new weights.
11 hours ago
Thinking output is already a relatively small part of the total input compared to raw code/text files.