11-19 更新 实时
首页/ 技术栈/ 正文

Pinecone 嵌入支持 text-embedding-3

技术栈 来源:Pinecone Blog 2026-11-11 00:00 0 阅读 机器翻译
Pinecone 嵌入支持 text-embedding-3

Pinecone 集成式嵌入现已支持 OpenAI 的 text-embedding-3-small 和 text-embedding-3-large,与此前随该功能一同推出的 Pinecone 托管模型并列。OpenAI 的 text-embedding-3 模型是我们收到需求最多的两个嵌入模型。团队现在可以直接基于它们运行语义搜索,无需构建和维护嵌入流水线。

在 Documents API 索引的某个文本字段上设置这两个模型中的任意一个,然后发送纯文本即可。Pinecone 会用该模型对该字段的每一次写入和每一次查询进行嵌入。无需管理 OpenAI 账户或密钥,Token 消耗会按 OpenAI 的标价显示在你的 Pinecone 账单上。

在 2026 年 12 月 31 日之前,Starter 和 Builder 套餐使用 OpenAI 模型时可获得其包含 Token 数的两倍:Starter 每个模型每月 10M,Builder 每个模型每月 20M。

让 Pinecone 来发起嵌入调用

使用 OpenAI 的嵌入模型意味着每一次写入和每一次查询都要调用 OpenAI 的 API。开发团队需要管理 OpenAI 账户、API 密钥和速率限制,并且要单独向 OpenAI 付费,与 Pinecone 分开结算。集成式嵌入把这些工作移入 Pinecone:

直接发送纯文本,不必先调用嵌入服务商。Pinecone 在一次调用中完成查询嵌入并执行搜索。

Pinecone 将每个文档的文本与其嵌入一起写入,因此全文搜索和语义搜索看到的始终是同一版本的文档。

Pinecone 通过 Microsoft Azure 提供 OpenAI 的模型服务,流量分布在多个 Azure 区域。当某个区域返回错误或变慢时,Pinecone 会在同一处理位置内的另一个区域自动重试请求,并将流量转移出去,直到该区域恢复,因此团队无需自行构建故障转移机制。

无需管理 OpenAI API key,也没有单独的 embedding 账单。Token 按 OpenAI 的标价计入 Pinecone 的账单。

如何在 Pinecone 中使用 OpenAI 的 text-embedding-3 模型

创建 Documents API 索引时,在文本字段的 embed 设置中指定模型。下面的 Python 示例使用 text-embedding-3-small 以 1,536 维对 body 字段做 embedding,并为 body 开启全文检索,因此同一个字段可同时服务于语义搜索和关键词搜索。

from pinecone import Pinecone pc = Pinecone(api_key="YOUR_API_KEY") pc.indexes.create( name="support-articles", schema={ "fields": { "body": { "type": "string", "full_text_search": {}, "embed": { "model": "text-embedding-3-small", "dimension": 1536, }, } } }, ) index = pc.index("support-articles") index.documents.upsert( namespace="articles", documents=[ {"_id": "kb-205", "body": "Refunds take five business days."}, ], )

搜索时,将查询以文本形式发送:

results = index.documents.search( namespace="articles", top_k=5, score_by=[ { "type": "embed", "fields": ["body"], "query": "how long do refunds take", } ], include_fields=["body"], )

如果省略 dimension,该字段将使用模型的完整维度。模型支持的任何维度都可用,因此你可以沿用现有索引已在使用的维度。

Pinecone 提供的 OpenAI 模型默认在全球范围内处理文本;当字段使用美国处理位置时,则仅在美国处理。这些模型在所有套餐中均可用,而美国处理位置需 Standard 及以上套餐。

这些模型可在无服务器的 Documents API 索引上运行。不支持 BYOC、基于 pod 的索引以及 Records API 索引。

控制台和 Python SDK 10.3.0 支持这些模型。Node.js 和 Go SDK 的支持即将推出。

现有的向量索引可以转换为 Documents API 索引,无需重新生成嵌入。在转换后的索引上添加集成嵌入的功能即将推出。在此之前,请在新的 Documents API 索引上使用这些模型,并把你的文本发送给它们。

Starter 套餐每月为每个 OpenAI 模型包含 5M Token,Builder 套餐每月为每个模型包含 10M。在 2026 年 12 月 31 日之前,两个套餐的额度均为这一数量的两倍。

阅读集成嵌入文档,了解如何在 Documents API 索引上设置 OpenAI 的 text-embedding-3 模型。在 2026 年 12 月 31 日之前,Starter 和 Builder 套餐可获得 OpenAI 模型所含 Token 的两倍额度:Starter 每个模型每月 10M,Builder 每个模型每月 20M。

这篇文章对你有帮助吗?

今天就开始构建知识丰富的 AI

免费创建你的第一个索引,待你准备好扩展规模时再按量付费。

译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。

查看英文原文(Pinecone integrated embedding now supports OpenAI's text-embedding-3 models)

Pinecone integrated embedding now supports OpenAI's text-embedding-3-small and text-embedding-3-large, alongside the Pinecone-hosted models it launched with. OpenAI's text-embedding-3 models are two of our most requested embedding models. Teams can now run semantic search on them without building and maintaining an embedding pipeline.

Set either model on a text field in a Documents API index and send plain text. Pinecone embeds every write and every query for that field with that model. There's no OpenAI account or key to manage, and the tokens appear on your Pinecone invoice at OpenAI's list price.

Through December 31, 2026, Starter and Builder plans get twice their included tokens for OpenAI's models: 10M a month on each model for Starter and 20M a month on each model for Builder.

Let Pinecone make the embedding calls

An application sends a document and a query to Pinecone as plain text, and Pinecone embeds both with OpenAI's text-embedding-3-small.

Using an OpenAI embedding model means calling OpenAI's API on every write and every query. Development teams manage the OpenAI account, API key, and rate limits, and pay OpenAI separately from Pinecone. Integrated embedding moves that work into Pinecone:

Send plain text instead of calling an embedding provider first. Pinecone embeds the query and runs the search in one call.

Pinecone writes each document's text and its embedding together, so full-text and semantic search always see the same version of each document.

Pinecone serves OpenAI's models from Microsoft Azure, with traffic spread across several Azure regions. When a region returns errors or slows down, Pinecone automatically retries requests in another region within the same processing location and shifts traffic away until the region recovers, so teams don't have to build their own failover.

There's no OpenAI API key to manage, and no separate bill for embedding. Tokens are billed on the Pinecone invoice at OpenAI's list price.

How to use OpenAI's text-embedding-3 models in Pinecone

Set the model in the embed setting of a text field when creating a Documents API index. The Python example below embeds the body field with text-embedding-3-small at 1,536 dimensions and turns on full-text search for body, so the same field serves semantic and keyword search.

from pinecone import Pinecone pc = Pinecone(api_key="YOUR_API_KEY") pc.indexes.create( name="support-articles", schema={ "fields": { "body": { "type": "string", "full_text_search": {}, "embed": { "model": "text-embedding-3-small", "dimension": 1536, }, } } }, ) index = pc.index("support-articles") index.documents.upsert( namespace="articles", documents=[ {"_id": "kb-205", "body": "Refunds take five business days."}, ], )

To search, send the query as text:

results = index.documents.search( namespace="articles", top_k=5, score_by=[ { "type": "embed", "fields": ["body"], "query": "how long do refunds take", } ], include_fields=["body"], )

If you leave out the dimension, the field uses the model's full size. Any size the model supports works, so you can keep the dimensions your existing indexes already use.

The OpenAI models Pinecone provides process text globally by default, or only in the United States when the field uses the US processing location. They're available on every plan, with the US processing location on the Standard plan and above.

The models work on serverless Documents API indexes. BYOC, pod-based, and Records API indexes aren't supported.

The console and Python SDK 10.3.0 support these models. Support in the Node.js and Go SDKs is coming soon.

Existing vector indexes can be converted to Documents API indexes without re-embedding. Adding integrated embedding to a converted index is coming soon. Until then, use these models on new Documents API indexes and send them your text.

Starter plan includes 5M tokens a month for each OpenAI model, and Builder plan includes 10M a month for each. Both plans get twice that through December 31, 2026.

Read the integrated embedding docs to learn how to set up OpenAI's text-embedding-3 models on a Documents API index. Through December 31, 2026, Starter and Builder plans get twice their included tokens for OpenAI's models: 10M a month on each model for Starter and 20M a month on each model for Builder.

Was this article helpful?

Start building knowledgeable AI today

Create your first index for free, then pay as you go when you're ready to scale.

原文为英文,正文由自动抓取并保留原文。

本文由采集器自动同步

原文发布于 Pinecone Blog(Pinecone integrated embedding now supports OpenAI's text-embedding-3 models),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。

阅读原文 ↗