Gemini 3.7 benchmarks against GDPVal-AA-V2 were 1525 in Aug blog post. However same model against same benchmark is 1482 in today's blog post of Gemini 3.8 release..
Do they make it intentionally to look previous model less superior than current models? or these are the real numbers when re-ran the benchmark..?
Do they make it intentionally to look previous model less superior than current models? or these are the real numbers when re-ran the benchmark..?