Yi-Large

Yi-Large is 01.AI's closed-source flagship model at 100B+ parameters.

On AlpacaEval 2.0 it trades first place with GPT-4; it excels at long-text understanding, deep summarization, and logical analysis, with a distinctive ability to capture deeper logic and philosophical nuance in summaries.

With a 16K context window (base version) and OpenAI-compatible APIs, it demonstrates flagship-grade depth and insight on analytical tasks, supporting the view that high performance can be achieved without relying solely on maximal scale.