🔥DeepSeek Input Cache Price Drop!
Effective immediately, the price for input cache hits across the ENTIRE DeepSeek API series is reduced to just 1/10th of the original price! Build more efficiently for less.
📌Reminder: The DeepSeek-V4-Pro 75% OFF promotion is still active until May 5th, 2026, 15:59 (UTC Time).
@thsottiaux Tibo, could you add a model selector to Codex on web and mobile?This would be really important for users like me whose computer is broken and who usually don’t carry a laptop when going out. If you can actually add this, I’ll keep subscribing to the Pro plan to support you guys
@nemumusitocha I wonder if you feel the same way: GPT-5.5 Pro is much faster and more creative than 5.4 Pro, but it seems like its accuracy and strictness have dropped a bit?
@CappaPython Also, 5.4 Pro seems replaced by a faster, distinct model—likely 5.5/5.5Pro. Opinions are mixed on whether it's better. It’s smarter to wait for the 5.5 series release to subscribe(plus, Pro models aren't available in Codex anyway)
@CappaPython 5.4Pro is powerful but incredibly slow.Basic tasks take at least 20 minutes,while complex ones can easily hit an hour—I once waited 120 minutes for a single prompt.Also,5.4 Pro’s coding isn't a huge leap over 5.4;its real strength lies in math and highly accurate data statistics
@gdb@gdb GPT-5.4 is excellent overall. I’d love an auto reasoning mode that adjusts to the question. In the new UI, manually switching the model, opening settings, and picking the reasoning level each time feels a bit cumbersome.
@sama When on earth are you finally going to release a new model? I hope you can hurry up and provide some satisfying models, whether it's called 5.3 or something else.