I mean, one thing people miss/forget is that we still don't have any LLM with the estimated compute budget of 4 that has failed to match it. Palm, Claude etc ar

I mean, one thing people miss/forget is that we still don't have any LLM with the estimated compute budget of 4 that has failed to match it. Palm, Claude etc are all lower. If Gemini fails to at least match 4 then we can start speculating on a secret sauce.