Curiosity · AI Model

Yi-Large

Yi-Large is the flagship hosted model from 01.AI — the Chinese AI lab founded by Kai-Fu Lee. Built alongside a family of open-weight Yi models on HuggingFace, Yi-Large is the closed, enterprise-tier offering: 32k context, bilingual Chinese/English excellence, and an API tuned for retrieval-augmented chat.

Model specs

Vendor
01.AI
Family
Yi
Released
2024-05
Context window
32,000 tokens
Modalities
text
Input price
$3/M tok
Output price
$3/M tok
Pricing as of
2026-04-20

Strengths

  • Strong bilingual (CN/EN) quality at a competitive price
  • Backed by experienced team (Kai-Fu Lee / Sinovation Ventures)
  • Companion open-weight Yi-34B available for self-hosting

Limitations

  • No vision or audio modalities
  • Smaller context than GLM-4 Plus (32k vs 128k)
  • Lags Qwen and DeepSeek on some 2025 reasoning benchmarks

Use cases

  • Chinese/English enterprise chatbots
  • RAG pipelines with bilingual knowledge bases
  • Content generation for Chinese domestic markets
  • Summarisation of long bilingual documents

Benchmarks

BenchmarkScoreAs of
MMLU~78%2026-04
C-Eval~82%2026-04
AlpacaEval 2.0~30%2026-04

Frequently asked questions

What is Yi-Large?

Yi-Large is 01.AI's flagship hosted LLM — a bilingual Chinese/English model tuned for enterprise chat and RAG. It is the closed, API-only counterpart to the open-weight Yi family on HuggingFace.

Who is behind 01.AI?

01.AI was founded in 2023 by Kai-Fu Lee, former head of Google China and founder of Sinovation Ventures. It is one of China's major LLM labs, alongside Zhipu, Moonshot, and Baichuan.

Sources

  1. 01.AI — Yi platform — accessed 2026-04-20
  2. 01.AI on HuggingFace — accessed 2026-04-20