Model

Yi-1.5 9B

OPEN

01.AI · Yi-1.5 family · released May 13, 2024

Popular mid-size Yi-1.5 variant, continually pretrained on 3.6T tokens; top performer among similarly sized open models at launch.

ReasoningCodingVisionFunction callingTool useAgentic

351.4

Elo · rank #350

Parameters

Active params

9B (dense)

Context

4K tokens

Architecture

Dense Llama-style decoder-only transformer

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

AGIEvalReasoning62.7%#7

best: OLMo 3-Think 32B · 88.2%

AlpacaEvalHuman preference22.9%#35

best: Tülu 2+DPO 70B · 95.1%

ARC-ChallengeReasoning61.5%#85

best: Llama 3.1 405B · 96.9%

Arena-HardHuman preference34.4%#49

best: Qwen3-235B-A22B (Non-Thinking) · 96.1%

BIG-Bench HardReasoning51.4%#97

best: ERNIE 4.5 300B-A47B · 94.3%

CMMLUKnowledge74.8%#17

best: Doubao-1.5-Pro · 90.9%

GPQA DiamondReasoning37.9%#236

best: GPT-6 Astra · 96.0%

GSM8KMath73.7%#96

best: Llama 3.1 405B · 96.8%

HumanEval+Coding34.1%#34

best: Mistral Small 3.2 (24B) · 92.9%

HumanEvalCoding41.4%#136

best: Claude Opus 4.5 · 99.4%

IFEvalReasoning29.4%#143

best: Gemma 4 26B A4B · 98.5%

MATH-500Math11.4%#191

best: GPT-5 · 99.4%

MBPP+Coding53.6%#33

best: Llama 3.1 405B · 88.6%

MBPPCoding61.1%#57

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

MMLU-ProKnowledge39.2%#154

best: Claude Fable 5 · 91.5%

MMLUKnowledge69.5%#148

best: OpenAI o3 · 92.9%

MT-BenchHuman preference8.2#42

best: Hunyuan-Large (A52B) · 9.4

Run it locally

VRAM @ Q4

6 GB

VRAM @ FP16

18 GB

Fits on (Q4)

RTX 3060 12GBRTX 4070 Ti 16GBRTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB

Throughput data unavailable.

Quantizations

GGUF Q4

Fine-tune it

Permissive

QLoRA7.7 GB1× RTX 3060 12GB

LoRA20.7 GB1× RTX 3090 24GB

Full fine-tune146.0 GB1× B200 192GB

QLoRA SFT on ~10k samples ≈ $5.27 (1× RTX 3060 12GB)

Plan a fine-tune →

Yi-1.5 family

Elo progression across releases

Yi-1.5 34BMay 2024491.7 Yi-1.5 9BMay 2024351.4

Resources

HF01-ai/Yi-1.5-9B GH01-ai/Yi-1.5

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