LANGUAGE MODEL 01.AI Last updated:

Yi-34B

01.AI's Frontier Open Bilingual Model

01.AI's first major open-weight model, released November 2023. Founded by Kai-Fu Lee (former Microsoft / Google China president, venture capitalist, well-known author of "AI Superpowers"), 01.AI's Yi-34B briefly held the top spot on Hugging Face's Open LLM Leaderboard — the first non-Western model to do so. Quality at 34B parameters approached Llama 2's 70B variant.

Try demo API Docs ↗
Cost
Free
Open weights — self-host
Context
4K
Up to 4,096 tokens
How are Intelligence, Speed & Cost bucketed?
Intelligence and Speed buckets are percentile ranks on Artificial Analysis. Cost buckets are fixed dollar thresholds keyed off output-token price ($/M out).
Intelligence
  • Top 1%≤ 1%
  • Top 5%≤ 5%
  • Top 10%≤ 10%
  • Good≤ 25%
  • Medium≤ 50%
  • Below avg> 50%
Speed
  • Top 1%≥ 345 tok/s
  • Top 5%≥ 237 tok/s
  • Top 10%≥ 196 tok/s
  • Good≥ 146 tok/s
  • Medium≥ 90 tok/s
  • Slow< 90 tok/s
Cost
  • Freeopen weights · self-host
  • Low< $1 / M out
  • Moderate$1–5 / M out
  • High≥ $5 / M out
Official ↗ GitHub ↗ Hugging Face ↗

Why it matters

Yi-34B's brief leaderboard SOTA was the moment the Hugging Face Open LLM community accepted that Chinese labs could ship at the open-frontier. The architectural-attribution controversy surfaced the harder question of how much credit Western open-source releases (Llama, Mistral) deserve in the Chinese open-LLM story.

Core Capabilities

Long Documents
Handles entire codebases, books, and multi-doc RAG.
Generative
Produces images, video, audio, or other media.

Context Window

4k tokens
≈ short doc
4k This model 本模型
32k Long docs 长文档
128k Books 整本书
400k Multi-doc 多文档
1M Codebase 整个代码库
10M

Availability

API
Not available
Product / App
Not available
Open Source
Released
Enterprise
—

Pricing Model

Free / self-host
Open weights — pay only for compute.
Self-host

Capability / Performance

Where this model sits relative to the middle 60% of models in the tree. All scores are 0–10 (higher is better).

Lower 20% Upper 80% This model
Context / memory
Context window size · log-scaled
6.0
9.0
1.0
Lower 20% 20th percentile — 20% of models score below this This model Where the current model lands Upper 80% 80th percentile — only 20% of models score above this Percentile boundaries are computed across every model in the tree that reports the underlying benchmark for each capability.

What it feels like

  • First Chinese open-weights model to top Hugging Face Open LLM Leaderboard for pre-trained models
  • Bilingual (English / Chinese) — trained on 3T tokens up to June 2023
  • Beat Falcon-180B and Llama 2 70B on multiple benchmarks while being smaller and bilingual
  • Ranked first on C-Eval Chinese benchmark; competitive with Claude on AlpacaEval
  • Earned 01.AI unicorn status within months of release — established the China open-weights playbook
  • Yi-34B-200K variant pushed to 200K context, ahead of most open peers in late 2023
Reviews: Yi technical report (arXiv) ↗ · Maginative — 01.AI reaches unicorn with Yi-34B ↗ · The Decoder — Yi beats Meta's Llama 2 ↗

Best use cases

  • Bilingual production deployments where Chinese performance matters
  • Self-hosted RAG and chat systems prioritising open-weights cost over closed-API quality
  • Distillation source for smaller bilingual models
  • Research on long-context tuning (Yi-34B-200K was an early demonstration)

Not ideal for

  • Frontier reasoning by 2024 — quickly surpassed by Qwen 2.5, DeepSeek V2/V3, Llama 3.x
  • Multimodal tasks (text-only at this generation)
  • Edge / single-GPU deployments at full 34B precision