TapTechNews May 30 news, Google at the I/O 2024 Developer Conference, announced that it will increase the Gemini 1.5 Pro context window to 2 million tokens. So just how powerful is the Gemini 1.5 Pro performance?
According to the comparison of the overall leaderboard released by LMSYSOrg, as measured by the ArenaElo system, Gemini-1.5-Pro-API-0514 and Gemini-Advanced-0514 are both close to GPT-4o.
The above two models also perform very well in Chinese, and Gemini 1.5 Pro is also very close in the hardprompts category.
TapTechNews note: The ArenaElo system measures the skills of large language models (LLMs) by having users anonymously vote to decide which model performs better in a random battle and updates users' ratings like the Elo system in chess, which is more objective overall.