Model details

Qwen3 Embedding 4B

Qwen · Qwen3-Embedding-4B

The training of the Qwen3 Embedding series models inherits the multi-stage training paradigm of the GTE-Qwen series, but has been deeply optimized for specific application scenarios. In the training process of the Embedding models, this series of models adopts a three-stage training architecture: the first stage conducts contrastive learning pre-training using ultra-large-scale weakly supervised data; the second stage conducts supervised training based on high-quality annotated data; finally, multiple candidate models are fused through a model fusion strategy to enhance overall performance. This staged training mechanism effectively balances the model's generalization ability and task adaptability.

能读 41K
Official price¥0.0668/ ¥0每百万 Token
中转最低6% of official¥0.0040 / ¥0
27家在卖

The best way to buy Qwen3 Embedding 4B

The best deal for each of three ways to buy. Tap a card to see every option of that kind.

Pay-as-you-go API · relay stationsFor coding and tools like Cursor and Claude Code
For using the web app and mobile app, no coding
Official subscriptionsFor a genuine official plan in your own name

Qwen3 Embedding 4B

The training of the Qwen3 Embedding series models inherits the multi-stage training paradigm of the GTE-Qwen series, but has been deeply optimized for specific application scenarios. In the training process of the Embedding models, this series of models adopts a three-stage training architecture: the first stage conducts contrastive learning pre-training using ultra-large-scale weakly supervised data; the second stage conducts supervised training based on high-quality annotated data; finally, multiple candidate models are fused through a model fusion strategy to enhance overall performance. This staged training mechanism effectively balances the model's generalization ability and task adaptability.

View intro source
Official price · Input¥0.0668
Official price · Output¥0.0000
Sellers27 家

The best way to buy Qwen3 Embedding 4B

The best deal for each of three ways to buy. Tap a card to see every option of that kind.

Pay-as-you-go API · relay stationsFor coding and tools like Cursor and Claude Code
For using the web app and mobile app, no coding
Official subscriptionsFor a genuine official plan in your own name
No reviews yet — write the first oneSee reviews ↓

模型介绍

The training of the Qwen3 Embedding series models inherits the multi-stage training paradigm of the GTE-Qwen series, but has been deeply optimized for specific application scenarios. In the training process of the Embedding models, this series of models adopts a three-stage training architecture: the first stage conducts contrastive learning pre-training using ultra-large-scale weakly supervised data; the second stage conducts supervised training based on high-quality annotated data; finally, multiple candidate models are fused through a model fusion strategy to enhance overall performance. This staged training mechanism effectively balances the model's generalization ability and task adaptability.

模型基础信息

厂商
通义千问
模型类型
向量与排序
一次能读多少
41K Token一次对话能塞进去的最大内容量,长文档、长代码要看这项
一次能写多少
40,960 Token单次回答的长度上限
能接收什么
文本可以发给它的内容类型
能产出什么
文本它能生成的内容类型
怎么计费
按 Token 计费官方按什么单位收钱
现在还能用吗
可以用 · 状态来源
这些资料是哪个地区的资料适用地区
全球按厂商全球公布的资料显示;国内没单独公布时,也用这一份。
它会做什么会做文本向量会写字

官方价格

折扣与价格评分按全球的官方价计算

计价项适用范围官方价格
输入全球0.0668 CNY每百万 Token
输出全球0 CNY每百万 Token
它还叫这些名字(6 条)
qwen-qwen3-embedding-4bQwen3-Embedding-4BQwen/QWEN3-EMBEDDING-4Baccounts/fireworks/models/QWEN3-EMBEDDING-4B通义千问-3-Embedding-4B通义千问-qwen3-embedding-4b

不同地方写法可能不一样,列在这里方便你对上号。名字太模糊、认不准的不会公开。

User reviews

No reviews yet. If you have used it, share the first honest review and help others avoid pitfalls.

Review this model

Rating
Optional
编辑器加载中…