Yi-Large
01.AI
SUPERSEDED. 01.AI's proprietary flagship model with 70B parameters, released May 2024. Chinese AI powerhouse trained from scratch with significant improvements over Yi-34B open-source model. Performs on par with GPT-4 and Claude 3 across benchmarks. Exceptional multilingual capabilities - excels in Spanish, Chinese, Japanese, German, and French per LMSYS multilingual leaderboard. Positioned first on AlpacaEval 2.0 shortly after launch. Trained on 3.1T tokens of English and Chinese using cascaded deduplication and quality filtering. Uses decoder-only transformer with pre-normalization, SwiGLU, RoPE, and Grouped Query Attention. 01.AI no longer documents this model; its published API line-up is led by yi-lightning. The fields that could not be read at a vendor URL have been removed rather than left asserted.
Strengths
- Performs on par with GPT-4 and Claude 3 across standard benchmarks
- Exceptional multilingual capabilities - Spanish, Chinese, Japanese, German, French
- Positioned #1 on AlpacaEval 2.0 shortly after launch
- Strong Chinese language performance - trained specifically for Chinese/English
- Trained on 3.1T tokens with cascaded deduplication and quality filtering
- Grouped Query Attention for efficient inference
- Represents Chinese AI competing globally with Western models
Caveats
- SUPERSEDED - 01.AI no longer documents this model; yi-lightning leads its current API line-up
- Context window, output limit, release date and pricing are not stated at any vendor URL and have been omitted
- 32K context window smaller than competitors (GPT-4o: 128K, Gemini: 1M-2M)
- Proprietary model - no open weights or self-hosting
- 4K max output tokens lower than newer flagship models
- Less ecosystem maturity than OpenAI/Anthropic in Western markets
- API availability may vary by region
- Vendor
- 01.AI
- License
- Proprietary
- Modalities
- text