it did jump up but in my experience Codex is still better than GH Copilot pricing for the same model(s).
I'm hoping after this price bump, they will start undercutting each other eventually and get it back to some more reasonable pricing model.
But the discounted prices only exist because of endless VC fundraising, right? I’m no expert, but I don’t reckon they’re sustainable without substantial outside funding. It seems akin to hoping uber would go back to charging like $8 to bring you to the airport in a limousine if competition compelled them to — going out of business would be the more likely outcome.
yes and no. Ai models become more efficient (I've seen a comparison somewhere). So even if the prices stay the same, usage could go down. Also, in the uber example, there's no free alternative, unlike ai models. I'm yet to try qwen and some others though, so can't vouch for them.
But that doesn’t seem like it would be them undercutting each other, per se. And to me it looks like most people are saying the number of tokens used per task just keeps going up, and on top of that, monthly plan pricing seems to be going away. I think we hit the summit of peak discount a few months ago, now they’re being forced to be more financially independent approaching their IPOs, and I just don’t see any factor feasibly forcing costs down after the subsidies go away. SpaceX had the rest of SpaceX to take the pressure off XAI being a money furnace, but what do OpenAI and Anthropic have?
+1 for codex (have moved to it from claude code; and am quite happy about it); but i do hope they keep pricing affordable and consistent