帖子
Google recently introduced Gemini 3.5 Flash and it surpasses Gemini 3.1 Pro across multiple areas. Coding performance reaches 76.2 percent while the prior version scores 70.3 percent. Real world agentic tasks show 1656 compared to 1314 in Elo rating. Scaled tool use hits 83.6 percent versus 78.2 percent. Consequently the new flash variant has made the pro model from the same company seem outdated. Observers describe this development as a major step forward. Improvements in tool utilization are evident with the 83.6 percent result on MCP Atlas. This outcome represents the progress m
3mo
62 赞2 踩2 转发1 评论
?
评论
还没有评论。来抢沙发吧！