Model

Yi-34B

OPEN

01.AI · Yi family · released Nov 2, 2023

Flagship bilingual EN/ZH base model that briefly topped the HF Open LLM Leaderboard in Nov 2023, matching GPT-3.5 on many benchmarks.

ReasoningCodingVisionFunction callingTool useAgentic

451.7

Elo · rank #318

Parameters

34B

Active params

34B (dense)

Context

4K tokens

Architecture

Dense decoder-only transformer (Llama-style, GQA, SwiGLU, RoPE with adjusted base frequency)

License

Apache 2.0

Languages

API price (in/out)

No hosted API

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

ARC-ChallengeReasoning64.6%#66

best: Llama 3.1 405B · 96.9%

BIG-Bench HardReasoning54.3%#87

best: ERNIE 4.5 300B-A47B · 94.3%

C-EvalKnowledge81.4%#32

best: Qwen3.6-Plus · 93.3%

CMMLUKnowledge83.7%#8

best: Doubao-1.5-Pro · 90.9%

DROPReasoning64.2#47

best: Hunyuan-T1 · 93.1

GSM8KMath67.2%#104

best: Llama 3.1 405B · 96.8%

HellaSwagReasoning85.7%#36

best: Claude 3 Opus · 95.4%

HumanEvalCoding23.2%#164

best: Claude Opus 4.5 · 99.4%

MATH-500Math14.4%#186

best: GPT-5 · 99.4%

MBPPCoding41.0%#84

best: Llama-3.3-Nemotron-Super-49B v1 (Reasoning On) · 91.3%

MMLUKnowledge76.3%#120

best: OpenAI o3 · 92.9%

TruthfulQAKnowledge56.2%#35

best: Phi-3.5-MoE (16x3.8B, 6.6B active) · 77.5%

WinoGrandeReasoning83.0%#19

best: PaLM 2 · 90.9%

Run it locally

VRAM @ Q4

22 GB

VRAM @ FP16

68.8 GB

Fits on (Q4)

RTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB

Throughput data unavailable.

Quantizations

GGUF Q4_K_M · AWQ · GPTQ

Yi family

Elo progression across releases

Yi-34BNov 2023451.7 Yi-6BNov 202342.9

Resources

HF01-ai/Yi-34B GH01-ai/Yi APIAvailable via 01.AI API

API price weights · each benchmark row carries its own source badge (see methodology)