4 ms·
It leads on arc-agi-1 with Gemini 3.0 Deep Think, which uses "tool calls" according to google's post, whereas regular Gemini 3.0 Pro doesn't use "tool calls" fo
by energy123 1y ago
It leads on arc-agi-1 with Gemini 3.0 Deep Think, which uses "tool calls" according to google's post, whereas regular Gemini 3.0 Pro doesn't use "tool calls" for the same benchmark. I am unsure how significant this difference is.