ERNIE-4.5-300B-A47B

聊天大模型

ERNIE-4.5-300B-A47B

Release date: 2025-06-30更新于: 2025-06-30 13:09:14974

Live demo GitHub Hugging Face Compare

Parameters

3000.0亿

Context length

128K

Chinese support

Supported

Reasoning ability

ERNIE-4.5-300B-A47B is an AI model published by 百度, released on 2025-06-30, for 聊天大模型, with 3000.0B parameters, and 128K tokens context length, requiring about 580GB storage, under the Apache 2.0 license.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

ERNIE-4.5-300B-A47B

Model basics

Reasoning traces

Not supported

Thinking modes

Thinking modes not supported

Context length

128K tokens

Max output length

65536 tokens

Model type

聊天大模型

Release date

2025-06-30

Model file size

580GB

MoE architecture

Yes

Total params / Active params

3000.0B / 470B

Knowledge cutoff

No data

ERNIE-4.5-300B-A47B

Open source & experience

Code license

Apache 2.0

Weights license

Apache 2.0- 免费商用授权

GitHub repo

https://github.com/PaddlePaddle/ERNIE

Hugging Face

https://huggingface.co/baidu/ERNIE-4.5-300B-A47B-PT

Live demo

https://yiyan.baidu.com/

ERNIE-4.5-300B-A47B

Official resources

Paper

Announcing the Open Source Release of the ERNIE 4.5 Model Family

DataLearnerAI blog

No blog post yet

ERNIE-4.5-300B-A47B

API details

API speed

3/5

💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.

Standard pricingStandard

Modality	Input	Output
Text	$0.55	$2.2

ERNIE-4.5-300B-A47B

Benchmark Results

ERNIE-4.5-300B-A47B currently shows benchmark results led by GSM8K (1 / 26, score 96.60), BBH (1 / 20, score 94.30), DROP (2 / 9, score 91.10). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.