주 콘텐츠로 건너뛰기
Azure

엔터프라이즈급 정보 검색 및 AI 애플리케이션 검색

Azure AI Search는 Gen AI 애플리케이션 내에서 RAG(검색 증강 생성)를 최적화하도록 설계된 정보 검색 및 검색 플랫폼입니다. 조직은 자체 데이터를 저장하고, 인덱싱하고, 검색하여 현재 정보를 AI 모델에 제공할 수 있습니다. 의미론적 순위, 벡터 및 하이브리드 검색을 비롯한 최첨단 기술로 가장 관련성이 높은 정보를 표시합니다.

가격 옵션 살펴보기

필터를 적용하여 요구 사항에 맞게 가격 옵션을 사용자 지정하세요.

가격은 예상값일 뿐이며 실제 가격 견적이 아닙니다. 실제 가격 책정은 Microsoft와 체결한 계약 유형, 구매 날짜 및 환율에 따라 다를 수 있습니다. 가격은 미국 달러를 기준으로 계산되며 전월 마지막 영업일 이전 영업일 기준 2일 동안 수집된 런던 종가 현물 환율을 사용하여 변환됩니다. 월말 전 영업일 기준 2일이 주요 시장의 공휴일인 경우에는 통상 영업일 기준 2일의 직전일을 기준금리 설정일로 합니다. 이 비율은 다가오는 달의 모든 거래에 적용됩니다. Azure 가격 책정 계산기에 로그인하여 Microsoft의 현재 프로그램/제안을 기반으로 가격 책정을 확인하세요. 가격 책정에 대한 자세한 내용을 알아보거나 가격 견적을 요청하려면 Azure 영업 전문가에게 문의하세요. Azure 가격 책정에 대한 자주 묻는 질문을 참조하세요.

서비스 가격 책정

무료 기본 표준 S1 표준 S2 표준 S3 스토리지 최적화 L1 스토리지 최적화 L2
스토리지1 50MB 15GB
(서비스당 최대 45GB)
160GB
(서비스당 최대 1.9 TB)
512GB
(서비스당 최대 6 TB)
1 TB
(서비스당 최대 12 TB)
2 TB
(서비스당 최대 24 TB)
4 TB
(서비스당 최대 48 TB)
서비스당 최대 인덱스 수 3 15 50 200 200 또는 1,000/고밀도 모드의 파티션 10 10
규모 확장 제한 N/A 서비스당 최대 9단위
(최대 3개의 파티션, 최대 3개의 복제본)
서비스당 최대 36단위
(최대 12개의 파티션, 최대 12개의 복제본)
서비스당 최대 36단위
(최대 12개의 파티션, 최대 12개의 복제본)
서비스당 최대 36단위
(최대 12개의 파티션, 최대 12개의 복제본)
고밀도 2모드에서 최대 3개 파티션
서비스당 최대 36단위
(최대 12개의 파티션, 최대 12개의 복제본)
서비스당 최대 36단위
(최대 12개의 파티션, 최대 12개의 복제본)
Compute Type is Default
SU(검색 단위)당 가격
$- $- $- $- $- $- $-
Compute Type is Confidential3
SU(검색 단위)당 가격
$- $- $- $- $- $- $-

1If you see lower storage in your service, you can upgrade.

2High density (HD) mode is an option available within the standard S3 service that allows a larger number of indexes to be created in a single service. For information on supported limits, refer here.

3Resources backed by confidential compute are available at additional cost. Some features are limited or unavailable, including semantic ranker, agentic retrieval, skillset execution, and vectorization. Learn more about confidential computing here.

에이전트 검색

Agentic retrieval incorporates conversation history and a bring-your-own-model (BYOM) integration into query planning, using an agentic framework to run and manage the RAG query pipeline.

How billing works for agentic retrieval:

  • Agentic retrieval billing has two parts: query planning and query execution (ranking)
  • Query planning is billed through the model used (i.e. Azure OpenAI GPT series) with input and output tokens. This is separate from agentic retrieval Azure AI Search costs.
  • Query execution (ranking) is part of the agentic retrieval process and billed through Azure AI Search.

제품/기능 가격 책정(1M 토큰)
에이전트 검색 First 50M tokens free per month.
$- per 1M additional tokens.

의미 순위매기기

Semantic ranker improves search relevance by finding content that is semantically similar to query terms. The service is only available for accounts on Basic, Standard tiers (S1, S2, and S3), and Storage-Optimized (L1 and L2) and has two pricing plans within those tiers. Use semantic ranker when you want to improve the quality of search results and optimize the user experience.

가격
의미 순위매기기 매월 처음 1000개 요청 무료. 1,000개 추가 요청당 $-입니다.

추가 기능

Custom Entity Lookup skill detects and labels documents containing user-defined words and phrases, and is available for all Basic, Standard, and Storage-Optimized tiers. It is ideal for defining and detecting specific entities in your data.

Document Cracking: Image Extraction retrieves content from a file during the built-in indexer document cracking phase and/or within the enrichment pipeline. Text extraction is free, while image extraction is billed during the initial document cracking step and/or when invoking the Document Extraction skill. Use this functionality when you have documents that contain images.

가격 정보
사용자 지정 엔터티 검색 기술 0~1M 텍스트 레코드 $-/텍스트 레코드 1,000개
1M~3M 텍스트 레코드 $-/텍스트 레코드 1,000개
3M~10M 텍스트 레코드 $-/텍스트 레코드 1,000개
10M+ 텍스트 레코드 $-/텍스트 레코드 1,000개
문서 크래킹: 이미지 추출 0~1M 이미지 $-/트랜잭션 1,000개
1M~5M 이미지 $-/트랜잭션 1,000개
5M+ 이미지 $-/트랜잭션 1,000개

비즈니스에 적합한 SKU를 선택하는 방법에 대한 지침

모든 비즈니스에는 성장하고 발전함에 따라 고유한 요구 사항이 있습니다. 조직에 적합한 비즈니스 모델로 시작하는 방법을 알아보려면 아래 지침을 살펴보세요.

  • SKU를 선택하기 전에 검색 서비스에 대한 요구 사항을 고려하는 것이 중요합니다. 다음은 질문해야 할 중요한 질문입니다.

    • 애플리케이션 사용자가 작성할 수 있는 쿼리 예제는 무엇인가요?
    • 어떤 종류의 정보를 검색할까요?
    • 사용자가 검색 결과를 필터링하거나 정렬할까요?
    • 필요한 인덱스 수는 얼마인가요?
    • 원시 데이터 파일 또는 Blob Storage의 크기는 어떻게 됩니까?
    • 나중에 새 문서/데이터를 수집하기 위해 헤드룸이 필요한가요?
  • Azure AI Search is offered in combinable search "units" that vary by storage and throughput. Selecting the optimal pricing tier for your solution is based on the following factors:

    • Storage requirements: This refers to the amount of data you need to store in your search service. Each tier has a storage limit per partition, and you can add up to 12 partitions on a search service. It's important that your service has enough space for your data and has headroom to grow if you plan on ingesting more data.

    Pro Tip: The storage size of your search index will typically be smaller than the size of your raw data, particularly for files like PDFs and images. The best way to assess the storage size required for your search index is to index a representative sample of documents into your search index in the Azure portal with the Import Data Wizard.


    • Queries per Second (QPS): The throughput of your search service, often measured in QPS, will depend on your data and the types of queries users will issue. It's best to run performance tests to understand the throughput a particular tier can handle for your scenario but we have performance benchmarks that show the throughput different customer scenarios. Check out our performance benchmarking guidance to see sample customer scenarios.

    For additional guidance, please see our documentation: Choosing a Pricing Tier.

  • Customers can further optimize their search experience with add-on services in the "Additional Features" table below. We also recommend customers read the following guidance on how to manage costs:

  • For additional help, please see our options for one-on-one guidance.

Azure 가격 책정 및 구매 옵션

Microsoft와 직접 연락하기

Azure 가격 책정을 살펴보세요. 클라우드 솔루션의 가격 책정을 이해하고 비용 최적화에 대해 알아보고 사용자 지정 제안을 요청하세요.

판매 전문가에게 문의하기

구매 방법 확인

Azure 웹 사이트, Microsoft 담당자 또는 Azure 파트너를 통해 Azure 서비스를 구매하세요.

옵션 살펴보기

추가 리소스

Azure AI Search

Azure AI Search 기능에 대해 자세히 알아보세요.

가격 계산기

요구 사항에 맞는 모든 Azure 제품을 사용하는 데 드는 월별 예상 비용을 산출해 보세요.

SLA

Azure AI Search에 대한 서비스 수준 계약을 검토하세요.

설명서

기술 자습서, 동영상, 추가 Azure AI Search 리소스를 검토하세요.

  • Microsoft Azure AI Search에서는 해당 시간에 사용한 단위 수를 기준으로 예측할 수 있는 시간당 고정 요금이 청구됩니다.
  • 1시간 이내에 검색된 마지막 검색 단위 수가 청구됩니다. 두 단위로 시작한 다음, 4개 단위로 확장한 다음 1시간 내에 모두 두 단위로 축소하면 두 단위에 대한 요금이 부과될 수 있습니다.
  • 중지 단추를 누르면 사용 중인 서비스 인스턴스로 들어오는 트래픽이 중지됩니다. 따라서 서비스는 여전히 실행되고 있으므로 시간당 요금이 계속 청구됩니다.
  • 사용량 또는 단위 활성 시간이 한 시간 미만인지에 관계없이 단위가 존재하는 시간마다 균일한 요금이 청구됩니다. 예를 들어 단위를 하나 만들고 5분 후 삭제하더라도 청구서에는 1 단위 시간에 대한 요금이 반영됩니다.
  • Azure AI Search 단위를 결합하여 추가 처리량 및 스토리지를 제공할 수 있습니다. 예를 들어, 1,500만 개 문서를 3,000만 개(추가 파티션)로 확장하려는 경우 단위 두 개를 구매하면 됩니다. 처리량을 늘리려면(추가 복제) 단위 두 개를 구입하면 됩니다. 스토리지와 처리량을 모두 늘리려는 경우 단위 네 개가 필요합니다(2개 복제 x 2개 파티션 = 4 검색 단위).
  • 무료는 개발자에게 Azure AI Search의 검색 기능과 구현을 테스트할 수 있는 샌드박스를 제공하기 위해 설계된 무료 버전의 Azure AI Search입니다. 프로덕션 작업에는 적합하지 않습니다. 기본, 표준 및 스토리지 최적화는 자체 관리형 SaaS(Search-as-a-Service) 솔루션의 이점을 활용하는 애플리케이션을 빌드하는 데 사용되는 옵션입니다. 표준에서는 스토리지와 애플리케이션 요구에 따라 확장되는 예측 가능한 처리량을 제공합니다. 스토리지 최적화 버전은 TB당 저렴한 가격으로 훨씬 더 많은 스토리지를 제공합니다. 많은 처리량을 요구하는 애플리케이션의 경우는azuresearch_contact@microsoft.com에 문의하세요.
  • Azure AI Search is available in the new Microsoft Azure portal. First, you must sign up for an Azure subscription, then you can add an Azure AI Search account to your Azure subscription via the gallery in the preview portal. Get more information.
  • Agentic retrieval autonomously decides the best sources and methods for retrieving information, enhancing response quality through reasoning and planning. Semantic ranker improves search relevance by understanding and leveraging the semantics of user queries. Learn more here.

판매 전문가에게 문의하여 Azure 가격을 알아보세요. 클라우드 솔루션의 가격을 파악하세요.

별도 비용이 없는 클라우드 서비스와 $200 크레딧을 사용하여 30일간 Azure를 체험해 보세요.

예상 비용에 추가되었습니다. 계산기에서 보려면 ‘v’를 누르세요.
무엇을 도와 드릴까요?