All tags
Topic: "model-ranking"
LMSys killed Model Versioning (gpt 4o 1120, gemini exp 1121)
gpt-4o-2024-11-20 gemini-exp-1121 deepseek-r1 openai google-deepmind anthropic deepseek mistral-ai model-release model-ranking open-source vision coding reasoning market-competition
AI News for 11/21/2024-11/22/2024 highlights the intense frontier lab race with OpenAI's gpt-4o-2024-11-20 and Google DeepMind's gemini-exp-1121 trading top spots on the Lmsys leaderboard. The trend of using date-based model identifiers instead of traditional versioning is noted across leading labs including Anthropic. DeepSeek R1 is gaining attention as a potent open-source alternative, especially in the context of the AI competition between China and the US. Gemini-Exp-1121 is praised for improvements in vision, coding, and reasoning, while MistralAI expands with a new Palo Alto office, signaling growth and hiring.