nvanlandschoot

Yeah, the progress is still incredibly impressive even if 15× is overstated. Curious to see how far it goes in the future.

“Haiku is faster than Opus” is fine as a simple statement. But if you’re going to say “15× faster” in a model card, it should be at similar accuracy. Otherwise you’re mostly comparing different settings, as opposed to a technological leap in model performance. It’s not technically wrong, it’s just not very useful and a bit misleading as a headline.

I think they modified the page. If you search for GPT-5.3-Codex-Spark, Google still has it indexed with 15x. Searching: GPT-5.3-Codex-Spark + "15x" will show all the downstream sites that picked up the claim.

Method: I used OpenAI’s published SWE-Bench Pro chart points and matched GPT-5.3-Codex-Spark to the baseline model at comparable accuracy levels by reasoning effort. At similar accuracy, the effective speedup is closer to ~1.37× rather than 15×.