3 ms·
Try gpt-5.3-codex-spark - it's 1000 TPS and from my experience more capable than 5.4 mini. If you have a subscription it's a different pool of usage.
by jasonjmcghee 4mo ago
Try gpt-5.3-codex-spark - it's 1000 TPS and from my experience more capable than 5.4 mini.
If you have a subscription it's a different pool of usage.
- small_model 4mo agoUsed it, very fast but tiny context window and doesn't have good reasoning. (good for quick simple code changes)
- beering 4mo agoAgreed, 1000tok/s just fills up the context window (which is big by 2004 standards) super fast. But seems like 5.3-spark was just a taste of what’s to come.
- taneq 4mo ago2004 standards? O.o
- partsch 4mo ago1904
- bogeholm 4mo agoBack when we were kids, we would get 0 tokens/sec _if we were lucky_
- mlinsey 4mo agoIn 2004, I took a class where we trained "language models" that were bigram word models, on an archive of a couple years of the Wall Street Journal. I remember someone who literally announced they were dropping the class to the whole room at the end of a lecture, saying "This isn't AI!!!"
- trollbridge 4mo agoMIMO 2.5 Pro ultraspeed has a 1M window. 1,000 tok/sec is great for planning since you can have a rapid conversation with a lot of turns.