{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-08-20T21:21:00.605Z","headline":"Mistral 推出 Agentic Search：多步检索提升 AI 系统复杂文档查询准确率","description":"Mistral 发布 Agentic Search，通过 search、open、navigate、read、grep 五工具的多步检索循环，让模型在长文档与多来源中查找、定位并验证信息。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","url":"https://www.aioga.com/news/cmt1pkwbj04bxroovzkfca5c7/","mainEntityOfPage":"https://www.aioga.com/news/cmt1pkwbj04bxroovzkfca5c7/","datePublished":"2026-08-20T16:02:07.000Z","dateModified":"2026-08-20T16:02:07.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://mistral.ai/news/agentic-search","https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7"],"canonicalUrl":"https://www.aioga.com/news/cmt1pkwbj04bxroovzkfca5c7/","directAnswer":{"@type":"Answer","text":"Mistral 发布 Agentic Search，采用 search、open、navigate、read、grep 五种工具构成多步检索循环，帮助 AI 系统在长文档和多来源数据中查找、定位并验证信息。","url":"https://www.aioga.com/news/cmt1pkwbj04bxroovzkfca5c7/","dateCreated":"2026-08-20T16:02:07.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"Mistral AI：News（网页 source article","url":"https://mistral.ai/news/agentic-search","datePublished":"2026-08-20T16:02:07.000Z","provider":{"@type":"Organization","name":"Mistral AI：News（网页","url":"https://mistral.ai/news/agentic-search"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","datePublished":"2026-08-20T16:02:07.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7"}}],"aggregationSource":"Mistral AI：News（网页","originalPublisher":{"name":"Mistral AI：News（网页","url":"https://mistral.ai/news/agentic-search"},"geoDeepAnswer":null,"article":{"id":"cmt1pkwbj04bxroovzkfca5c7","slug":"cmt1pkwbj04bxroovzkfca5c7","url":"https://www.aioga.com/news/cmt1pkwbj04bxroovzkfca5c7/","title":"Mistral 推出 Agentic Search：多步检索提升 AI 系统复杂文档查询准确率","title_en":"","summary":"Mistral 发布 Agentic Search，通过 search、open、navigate、read、grep 五工具的多步检索循环，让模型在长文档与多来源中查找、定位并验证信息。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","source":"Mistral AI：News（网页","sourceUrl":"https://mistral.ai/news/agentic-search","aiHotUrl":"https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","publishedAt":"2026-08-20T16:02:07.000Z","category":"行业动态","score":72,"selected":true,"articleBody":["Build, test, and run AI agents and apps.","Train, align, and evaluate custom AI models.","Coding agents in the terminal, IDE, and background.","Frontier-scale infrastructure for training and inference.","Agentic Search. More accurate and efficient results from your AI systems.","In-region inference, open models, and new European infrastructure for sovereign AI.","Mistral Agentic Search delivers more accurate search results while reducing turns, token use, and latency against FinanceBench and OfficeQA Pro benchmarks. Agentic Search is the retrieval layer that enables AI systems to navigate, read, and verify information inside even the most complex documents. Available through Mistral Search Toolkit and Libraries.","Mistral Agentic Search helps enterprises get better results from their AI systems by letting models search and navigate their organization’s most complex data and documents. Agentic Search introduces a multi-step retrieval loop for finding, inspecting, and verifying information across data sources, wherever it is stored. Agentic Search is available through Mistral Search Toolkit ：https://docs.mistral.ai/studio/search-toolkit, built into Libraries ：https://docs.mistral.ai/studio/libraries in both Studio ：https://mistral.ai/products/studio/ and Vibe ：https://mistral.ai/products/vibe/, and gives you:","Support for sensitive domain-specific data . Mistral’s portable and open tooling helps you unlock value from your data without crossing your isolation boundaries in the cloud or on-premises.","Improved search results. Your models can search and navigate your data beyond retrieved chunks–inside long, dense documents or across multiple sources.","Access to existing indexes. Agentic Search builds on your existing search index using five tools: search , open , navigate , read , and grep .","Higher accuracy. Agentic Search delivers to 3x correctness on financial filings, from 26.7% to 86%, based on FinanceBench. On table-heavy, multi-doc questions of the OfficeQA Pro benchmark, we measure a +45.6 point gain (6.3% to 51.9%).","Lower latency and token use. Targeted navigation enables Agentic Search to reduce p90 latency up to 39.6% . Fewer repeated searches reduce token consumption by up to one-third.","Competitive edge is built upon years of real-world operations–your data, your processes, and your domain expertise. Proprietary knowledge is both critical to your success and highly confidential, meaning it lives behind isolation boundaries, segmented deployments, and self-hosted platforms. It accumulates in financial filings, legal contracts, internal resources, and government records–long, dense documents that traditional search methods can’t navigate effectively.","Agents that learn and improve continuously can help you compound your competitive advantage, but these agents are often separated from confidential data and proprietary knowledge for security reasons. Getting real impact from AI means pairing frontier reasoning with retrieval tools that can safely reach your most sensitive material.","Traditional, one-shot RAG retrieves a fixed set of text chunks and asks a model to answer in a single pass. This works when the answer appears in one of the top results, but falters when the model must navigate a long report, follow references, compare multiple documents, or verify the underlying evidence.","The limitation is more pronounced on dense, complex data and documents. The information needed to answer a question may be spread across documents or buried in a particular table, footnote, or clause. One-shot RAG-based search fails to use the full power of frontier AI and to provide reliable answers for three reasons:","Retrieval without reasoning: The model must answer from the chunks selected during the initial retrieval, even when they are incomplete or not relevant. It cannot decide that it needs a different document, another section, or more context before responding, which limits the impact of the model’s reasoning.","Chunk-level limit: Critical data is often held in complex multi-modal documents. When asked, “ What was the company’s effective tax rate in Q3?” an index may find the correct document but cannot open it, navigate to the table, read the surrounding context, or verify the answer.","No iteration: Many questions need more than one retrieval pass to get the correct answer. The model may need to refine its search, inspect a promising document, follow a reference, compare multiple sources, keep track of what it has seen, and try a new route when the first results are insufficient. One-shot RAG provides no way to take these next steps.","Without Agentic Search (one-shot retrieval)","Using specifically only the reported values for all individual calendar months in 1953, what is the total sum of these values of expenditures for U.S. national defense and associated activities (in millions of nominal dollars)?","Trajectory 1 tool_call (search only)","search(\"national defense expenditures monthly 1953\") → 10 hits: a scatter of monthly bulletins (Table 3), each framed fiscal-year, covering only part of 1953.","I found January–June 1953 data. But I need July–December 1953 monthly values to compute an answer.","Trajectory 3 tool_calls (2× search → read)","search(\"national defense expenditures monthly 1953\") → per-month bulletins (partial year)","search(\"…1953 November December 1954 to date\") → surfaces treasury_bulletin_1954_02.pdf p.15 (Table 3, all 12 months of 1953)","read(treasury_bulletin_1954_02.pdf, p.15) → pulls the complete Table 3","Mistral Search Toolkit ：https://docs.mistral.ai/en/studio/search-toolkit provides open modules for ingesting, embedding, and indexing critical and complex data in the cloud or on-premises. Agentic Search builds on this index by giving the model five tools that resemble familiar file-system operations:","search finds relevant documents across the corpus using the existing index.","navigate moves to a page, section, or region within it.","read retrieves the content at that location.","grep finds a pattern within an open document.","Rather than answering only from the initial top- k results, the model can inspect what it finds, refine its search, open relevant documents, navigate to specific sections, and read the source material before answering. The index identifies likely sources; Agentic Search determines what to inspect within and across them.","These tools do not require fine-tuning or model-specific training. As models get better at reasoning and tool use, retrievals get better without infrastructure changes. This is a key property: retrieval quality scales with model capability instead of being capped by your chunking strategy.","Long documents. Filings, contracts, manuals, technical specifications, and reports where the answer may appear on a particular page or in a specific table, clause, figure, or footnote.","Questions across multiple sources. Research that requires the model to find, compare, or reconcile evidence from several documents before reaching an answer.","Answers that must be verified. Financial figures, legal clauses, regulatory references, and operational data, where the response can be referenced in a stable and specific document location.","Tables and structured documents. Financial statements, government records, and scanned PDFs where meaning depends on rows, columns, page position, or surrounding context–not narrative text alone.","Direct lookups. Short, clean documents where the answer is likely to appear in one of the first retrieved chunks.","High-volume search. Keyword or semantic lookups that need to return relevant passages without reasoning over or navigating through them.","Simple, predictable questions. Use cases where the likely source and location of the answer are known in advance and additional retrieval steps are unlikely to improve the result.","One-shot RAG is often sufficient for these searches. Add Agentic Search when questions require the model to move beyond the initial results and investigate the source material. A well-configured index remains the right foundation in both cases.","We benchmarked Agentic Search on two industry-standard evaluations, using the out-of-the-box Mistral Search Toolkit stack: default chunking, default ranking, no tuning. These results are floors, not ceilings, meaning you can further improve result quality with use-case-specific tuning.","With these benchmarks, we tested two models using the Mistral Search Toolkit: Mistral Medium 3.5 (MM 3.5) and Z.ai GLM-5.2 (GLM-5.2), showcasing performance of a smaller model (MM 3.5) and a larger model (GLM-5.2).","Benchmark results are consistent: the agentic loop delivers substantive quality improvements and navigation tools increase accuracy while reducing wasted tokens, turns, and latency. We observe the same performance patterns across first- and third-party models, which indicates that Agentic Search is model-agnostic, and that search quality should improve with new models.","FinanceBench (Islam et al., 2023) tests financial question-answering over 368 SEC filings (10-K / 10-Q / 8-K), averaging ~147 pages each, ~53,900 pages total: long, table-heavy financial documents. Answers scored by an LLM judge calibrated against human labels.","The search-only Agentic loop is the biggest quality lever. Moving from one-shot RAG to a search-only loop lifts accuracy by +47.3pp for MM 3.5 and +52.6pp for GLM-5.2–a ~3x improvement for both models. Because models can search iteratively, they can recover from weak first results, refine queries, and use the index as an active tool.","Navigation adds accuracy. Adding open, navigate, read, and grep lifts accuracy again ( +8.7pp for MM 3.5, +6.7pp for GLM-5.2). This means a targeted drill-in search beats repeated broad search in complex documents.","Token and performance efficiency improve with better retrieval tools. The full loop with Navigation answers more questions correctly while using fewer tokens than the search-only loop (MM 3.5: -23.9% token usage , GLM-5.2: -33.7% ). The retrieval tools are not additional overhead–they replace wasted search retries with precise navigation.","Latency goes down where it matters. Across FinanceBench, adding navigation retrieval tools improves latency: p90 drops 255s → 154s and mean latency drops 108s → 71s . In general, we see the search-only loops conduct repeated broad searches, while navigation helps the model identify evidence more quickly.","OfficeQA Pro is a verifiable numeric benchmark over historical U.S. Treasury Bulletins: scanned, table-heavy government-finance PDFs across a 696-document, ~89,000-page corpus. We report the first pass for the 133-question \"pro\" subset.","Agentic Search and the Agentic loop + Navigation are successful against a harder, verifiable benchmark. OfficeQA Pro has numeric answers, scanned PDFs, and deep table lookups. Even here, the full agentic loop lifts accuracy materially from one-shot RAG, reaching 51.9% for GLM-5.2 ( +45.6pp ) and increasing +27.1pp for MM 3.5.","Navigation improves quality while cutting waste. Using the full loop (Agentic loop + Navigation) improves accuracy by up to 35.6% ( +7.5pp, MM 3.5; +8.3pp, 19.0% GLM-5.2), while reducing token consumption. Turns declined by up to 7.0% (MM 3.5, 2.3% GLM-5.2).","The harder the benchmark, the more important the retrieval loop becomes. OfficeQA Pro is built around numeric answers in scanned, table-heavy documents. One-shot RAG barely gets started, while the agentic loop allows the model to search iteratively, inspect evidence, and deliver substantial accuracy improvements.","The tooling stack drives substantial impact on document intelligence and search performance. Per Kimi research ：https://www.kimi.ai/blog/kimi-k3, GLM-5.2 scores 41.4% on OfficeQA Pro with the Claude Code harness, compared with 51.9% on the Mistral harness–+10.5pp on the same underlying model.","Learn more about Agentic Search in the documentation ：https://docs.mistral.ai/studio/search/agentic-search. You can get started across cloud and on-premises deployments using either:","Mistral Search Toolkit ：https://docs.mistral.ai/studio/search/search-toolkit. Integrate Agentic Search into your own agents, workflows, and customer deployments.","Libraries ：https://docs.mistral.ai/studio/libraries. Use Agentic Search out-of-the-box in Studio and Vibe, without building the retrieval system yourself."],"articleImages":[{"sourceUrl":"https://mistral.ai/cms-media/api/media/file/2a1ffaf3-f171-460b-be1b-fc734aa776aa.svg","alt":"2a1ffaf3-f171-460b-be1b-fc734aa776aa","afterParagraph":3,"url":"/media/articles/cmt1pkwbj04bxroovzkfca5c7/e32e6b666122f9f7.jpg"},{"sourceUrl":"https://mistral.ai/cms-media/api/media/file/icon-m-microphone.svg","alt":"","afterParagraph":3,"url":"/media/articles/cmt1pkwbj04bxroovzkfca5c7/c42626ebf983a308.jpg"}],"mediaStatus":"ok","articleBodyZh":["构建、测试并运行 AI 代理和应用程序。","训练、对齐并评估自定义 AI 模型。","在终端、IDE 和后台编写代理代码。","用于训练和推理的前沿规模基础设施。","代理搜索。从您的 AI 系统获得更准确和高效的结果。","区域内推理、开放模型以及用于主权 AI 的新欧洲基础设施。","Mistral 代理搜索在 FinanceBench 和 OfficeQA Pro 基准上提供更准确的搜索结果，同时减少回合次数、令牌使用量和延迟。代理搜索是检索层，使 AI 系统能够在即使是最复杂的文档中导航、阅读和验证信息。可通过 Mistral 搜索工具包和库获得。","Mistral 代理搜索帮助企业通过让模型搜索和导航其组织中最复杂的数据和文档，从 AI 系统中获得更好的结果。代理搜索引入多步骤检索循环，以便在数据源中查找、检查和验证信息，无论其存储在哪里。代理搜索可通过 Mistral 搜索工具包使用：https://docs.mistral.ai/studio/search-toolkit，在库中集成使用：https://docs.mistral.ai/studio/libraries，并在 Studio：https://mistral.ai/products/studio/ 和 Vibe：https://mistral.ai/products/vibe/ 中提供，功能包括：","支持敏感的特定领域数据。Mistral 的可移植和开放工具帮助您在不跨越云或本地隔离边界的情况下释放数据价值。","改进的搜索结果。您的模型可以搜索并导航数据，不仅限于检索块——可在长且密集的文档中或多个来源之间进行检索。","访问现有索引。代理搜索通过五种工具建立在现有搜索索引的基础上：搜索、开放、导航、阅读和 grep。","更高的准确性。在金融报表上，基于 FinanceBench，代理搜索将正确率提高至 3 倍，从 26.7% 提升至 86%。在 OfficeQA Pro 基准的表格密集、多文档问题上，我们测量到 +45.6 个百分点的增幅（从 6.3% 提升到 51.9%）。","更低的延迟和令牌使用量。针对性导航使代理搜索将 P90 延迟降低最多 39.6%。减少重复搜索可将令牌消耗减少多达三分之一。","竞争优势建立在多年的实际操作之上——你的数据、你的流程以及你的领域专业知识。专有知识对于你的成功至关重要，同时高度保密，这意味着它存在于隔离边界、分段部署和自托管平台之中。它积累在财务报表、法律合同、内部资源和政府记录中——这些都是冗长、密集的文档，传统搜索方法无法有效浏览。","能够持续学习和改进的智能代理可以帮助你累积竞争优势，但出于安全原因，这些代理通常与机密数据和专有知识隔离。要从人工智能中获得实际影响，需要将前沿推理与能够安全访问最敏感材料的检索工具结合。","传统的一次性RAG检索获取固定的文本块，并让模型在一次处理过程中给出答案。当答案出现在前几个结果中时，这种方法可行，但当模型必须浏览长篇报告、跟踪参考资料、比较多份文件或验证基础证据时，就会失败。","在密集且复杂的数据和文档上，这种限制更加明显。回答问题所需的信息可能分散在多个文档中，或埋藏在特定表格、脚注或条款中。基于一次性RAG的搜索无法充分利用前沿AI的全部能力，也无法提供可靠答案，原因有三：","无推理的检索：模型必须基于初始检索中选择的文本块回答问题，即使这些文本块不完整或无关。它无法决定在回答之前需要其他文档、其他部分或更多上下文，从而限制了模型推理的影响力。","文本块级限制：关键数据通常存在于复杂的多模态文档中。比如被问及“公司第三季度的有效税率是多少？”时，索引可能找到正确的文档，但无法打开文档、定位表格、阅读周围上下文或验证答案。","无迭代：许多问题需要多次检索才能得到正确答案。模型可能需要优化其搜索、检查有希望的文档、跟随引用、比较多个来源、跟踪已查看的信息，并在第一次结果不足时尝试新的途径。一站式RAG无法执行这些后续步骤。","没有代理式搜索（一站式检索）","仅使用1953年所有单独日历月份的报告值，美国国防及相关活动的支出总额（以百万名义美元计）是多少？","轨迹1 工具调用（仅搜索）","search(\"national defense expenditures monthly 1953\") → 10条命中：一些按月公告的零散信息（表3），每个覆盖财政年度的一部分，仅涵盖1953年的部分时间。","我找到了1953年1月至6月的数据。但我需要1953年7月至12月的月度数据才能计算答案。","轨迹 3 工具调用（2×搜索 → 阅读）","搜索(\"1953年国防开支每月\") → 每月公告（部分年份）","search(\"…1953年11月到1954年12月至今\") → 显示 treasury_bulletin_1954_02.pdf 第15页（表3，1953年的全部12个月）","read(treasury_bulletin_1954_02.pdf, p.15) → 提取完整的表3","Mistral Search Toolkit：https://docs.mistral.ai/en/studio/search-toolkit 提供用于在云端或本地摄取、嵌入和索引关键复杂数据的开放模块。代理式搜索基于该索引，为模型提供五种类似熟悉文件系统操作的工具：","search 使用现有索引在语料库中查找相关文档。","navigate 移动到页面、章节或其中的区域。","read 会检索该位置的内容。","grep 在打开的文档中查找匹配的模式。","模型不仅仅从初始的前k个结果回答问题，还可以检查找到的内容、优化搜索、打开相关文档、导航到特定部分并读取源材料后再回答。索引确定可能的来源；代理式搜索确定在这些来源内及之间需要检查的内容。","这些工具不需要微调或特定模型的训练。随着模型在推理和工具使用方面的能力提升，检索效果会在不改变基础设施的情况下提升。这是一个关键特性：检索质量随模型能力提升而提升，而不会被分块策略限制。","长文档。文件、合同、手册、技术规格和报告，其中答案可能出现在特定页面、特定表格、条款、图表或脚注中。","跨多个来源的问题。需要模型在得出答案之前，从多份文档中查找、比较或协调证据的研究。","必须验证的答案。财务数据、法律条款、监管参考和运营数据，其回答可以在一个稳定和特定的文档位置中被引用。","表格和结构化文档。财务报表、政府记录和扫描PDF，其中含义依赖于行、列、页面位置或周围上下文，而不仅仅是叙述文本。","直接查找。简短且干净的文档，答案可能出现在检索到的前几个分块中。","高量搜索。关键词或语义检索，需要返回相关段落，而不需要对其进行推理或浏览。","简单、可预测的问题。使用场景中答案的来源和位置事先已知，并且额外的检索步骤不太可能改善结果。","一次性RAG通常足以满足这些搜索需求。当问题需要模型超出初始结果并调查源材料时，添加Agentic搜索。配置良好的索引在两种情况下仍然是正确的基础。","我们使用开箱即用的Mistral搜索工具包堆栈对两个行业标准评估进行了Agentic搜索基准测试：默认分块、默认排序、无调优。这些结果是下限，而非上限，意味着你可以通过特定使用场景的调优进一步提升结果质量。","通过这些基准，我们使用Mistral搜索工具包测试了两个模型：Mistral Medium 3.5 (MM 3.5) 和 Z.ai GLM-5.2 (GLM-5.2)，展示了一个小模型(MM 3.5)和一个大型模型(GLM-5.2)的性能。","基准测试结果一致：代理循环提供了实质性的质量提升，导航工具在提高准确性的同时减少了无效的令牌、对话轮次和延迟。我们在一方和第三方模型中观察到相同的性能模式，这表明代理搜索（Agentic Search）与模型无关，并且随着新模型的出现，搜索质量应会提高。","FinanceBench（Islam 等, 2023）在368份美国证券交易委员会文件（10-K / 10-Q / 8-K）上测试财务问答，每份文件平均约147页，总计约53,900页：长篇且以表格为主的财务文件。答案由与人工标签校准的LLM评审进行评分。","仅搜索的代理循环是质量提升的最大杠杆。从一次性RAG（检索增强生成）到仅搜索循环，MM 3.5的准确率提升+47.3个百分点，GLM-5.2提升+52.6个百分点——对两种模型来说约为3倍的改进。因为模型可以迭代搜索，它们可以从初步结果的不足中恢复，改进查询，并将索引用作主动工具。","导航提升准确率。增加打开、导航、阅读和搜索（grep）功能再次提高准确率（MM 3.5 +8.7个百分点，GLM-5.2 +6.7个百分点）。这意味着在复杂文档中，针对性的深入搜索比反复的广泛搜索更有效。","更好的检索工具提高了令牌和性能效率。带导航的完整循环在使用比仅搜索循环更少的令牌的同时，回答更多问题正确（MM 3.5: 令牌使用-23.9%，GLM-5.2: -33.7%）。检索工具不是额外负担——它们用精准的导航替代了无效的搜索重试。","延迟在关键处下降。在FinanceBench中，添加导航检索工具提高了延迟表现：90百分位延迟从255秒降至154秒，平均延迟从108秒降至71秒。总体来看，我们观察到仅搜索循环会进行反复的广泛搜索，而导航帮助模型更快识别证据。","OfficeQA Pro是在历史美国财政部公报上的可验证数值基准：跨696份文档、约89,000页的扫描版、表格密集型政府财务PDF。我们报告了133个“Pro”子集问题的首次测试结果。","代理搜索和代理循环+导航在更严格、可验证的基准测试中表现出成功。OfficeQA Pro 包含数值答案、扫描 PDF 和复杂表格查找。即便在这种情况下，完整的代理循环也显著提升了准确性，相较于一次性 RAG，GLM-5.2 达到 51.9%（+45.6 个百分点），MM 3.5 提升了 +27.1 个百分点。","导航在提升质量的同时减少了浪费。使用完整循环（代理循环+导航）可以提高准确率最多 35.6%（MM 3.5 +7.5 个百分点；GLM-5.2 19.0% +8.3 个百分点），同时减少了 token 消耗。对话轮数下降最多 7.0%（MM 3.5，GLM-5.2 2.3%）。","基准越严格，检索循环的重要性越高。OfficeQA Pro 构建在扫描、表格密集文档中的数值答案上。一-shot RAG 几乎无法起步，而代理循环允许模型迭代搜索、检查证据并显著提高准确性。","工具堆栈对文档智能和搜索性能产生了显著影响。根据 Kimi 研究：https://www.kimi.ai/blog/kimi-k3，使用 Claude Code 束缚时，GLM-5.2 在 OfficeQA Pro 上得分 41.4%，而在 Mistral 束缚上得分为 51.9%——同一模型提升 +10.5 个百分点。","想了解更多关于代理搜索的信息，请查阅文档：https://docs.mistral.ai/studio/search/agentic-search。您可以在云端和本地部署中使用以下任一方式开始使用：","Mistral 搜索工具包：https://docs.mistral.ai/studio/search/search-toolkit。将 Agentic 搜索集成到您自己的代理、工作流程和客户部署中。","库：https://docs.mistral.ai/studio/libraries。在 Studio 和 Vibe 中开箱即用地使用代理搜索，无需自行构建检索系统。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Mistral 发布 Agentic Search，采用 search、open、navigate、read、grep 五种工具构成多步检索循环，帮助 AI 系统在长文档和多来源数据中查找、定位并验证信息。","background":"该能力被定位为 AI 系统的检索层，可在复杂文档和企业数据中执行搜索、导航、阅读与验证。材料显示，Agentic Search 已通过 Mistral Search Toolkit 与 Libraries 提供。","viewpoint":"Aioga 判断，多步检索有望提升复杂文档问答的事实核验能力。Mistral 称其在 FinanceBench 与 OfficeQA Pro 基准上减少了交互轮次、令牌使用量和延迟，但材料未提供具体数据。","implications":"值得关注的是，企业级 AI 应用的检索重点正从单次搜索转向跨来源的连续定位与验证。实际效果仍需结合企业数据结构、文档质量和具体部署场景评估。","nextStep":"建议进一步核对 FinanceBench 与 OfficeQA Pro 的具体测试结果，并评估该检索循环在企业复杂文档、跨数据源查询和实际延迟控制中的表现。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-08-20T16:24:08.259Z","sourceHash":"c9a2064b1282c465","review":{"approved":true,"groundedness":96,"clarity":92,"duplicationRisk":12,"blockingIssues":[],"notes":[]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","Mistral AI：News（网页）"],"translations":{"zh-CN":{"title":"Mistral 推出 Agentic Search：多步检索提升 AI 系统复杂文档查询准确率","summary":"Mistral 发布 Agentic Search，通过 search、open、navigate、read、grep 五工具的多步检索循环，让模型在长文档与多来源中查找、定位并验证信息。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral 推出 Agentic Search：多步检索提升 AI 系统复杂文档查询准确率 - Aioga AI资讯","description":"Mistral 发布 Agentic Search，通过 search、open、navigate、read、grep 五工具的多步检索循环，让模型在长文档与多来源中查找、定位并验证信息。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","url":"https://www.aioga.com/news/cmt1pkwbj04bxroovzkfca5c7/","articleBody":["构建、测试并运行 AI 代理和应用程序。","训练、对齐并评估自定义 AI 模型。","在终端、IDE 和后台编写代理代码。","用于训练和推理的前沿规模基础设施。","代理搜索。从您的 AI 系统获得更准确和高效的结果。","区域内推理、开放模型以及用于主权 AI 的新欧洲基础设施。","Mistral 代理搜索在 FinanceBench 和 OfficeQA Pro 基准上提供更准确的搜索结果，同时减少回合次数、令牌使用量和延迟。代理搜索是检索层，使 AI 系统能够在即使是最复杂的文档中导航、阅读和验证信息。可通过 Mistral 搜索工具包和库获得。","Mistral 代理搜索帮助企业通过让模型搜索和导航其组织中最复杂的数据和文档，从 AI 系统中获得更好的结果。代理搜索引入多步骤检索循环，以便在数据源中查找、检查和验证信息，无论其存储在哪里。代理搜索可通过 Mistral 搜索工具包使用：https://docs.mistral.ai/studio/search-toolkit，在库中集成使用：https://docs.mistral.ai/studio/libraries，并在 Studio：https://mistral.ai/products/studio/ 和 Vibe：https://mistral.ai/products/vibe/ 中提供，功能包括：","支持敏感的特定领域数据。Mistral 的可移植和开放工具帮助您在不跨越云或本地隔离边界的情况下释放数据价值。","改进的搜索结果。您的模型可以搜索并导航数据，不仅限于检索块——可在长且密集的文档中或多个来源之间进行检索。","访问现有索引。代理搜索通过五种工具建立在现有搜索索引的基础上：搜索、开放、导航、阅读和 grep。","更高的准确性。在金融报表上，基于 FinanceBench，代理搜索将正确率提高至 3 倍，从 26.7% 提升至 86%。在 OfficeQA Pro 基准的表格密集、多文档问题上，我们测量到 +45.6 个百分点的增幅（从 6.3% 提升到 51.9%）。","更低的延迟和令牌使用量。针对性导航使代理搜索将 P90 延迟降低最多 39.6%。减少重复搜索可将令牌消耗减少多达三分之一。","竞争优势建立在多年的实际操作之上——你的数据、你的流程以及你的领域专业知识。专有知识对于你的成功至关重要，同时高度保密，这意味着它存在于隔离边界、分段部署和自托管平台之中。它积累在财务报表、法律合同、内部资源和政府记录中——这些都是冗长、密集的文档，传统搜索方法无法有效浏览。","能够持续学习和改进的智能代理可以帮助你累积竞争优势，但出于安全原因，这些代理通常与机密数据和专有知识隔离。要从人工智能中获得实际影响，需要将前沿推理与能够安全访问最敏感材料的检索工具结合。","传统的一次性RAG检索获取固定的文本块，并让模型在一次处理过程中给出答案。当答案出现在前几个结果中时，这种方法可行，但当模型必须浏览长篇报告、跟踪参考资料、比较多份文件或验证基础证据时，就会失败。","在密集且复杂的数据和文档上，这种限制更加明显。回答问题所需的信息可能分散在多个文档中，或埋藏在特定表格、脚注或条款中。基于一次性RAG的搜索无法充分利用前沿AI的全部能力，也无法提供可靠答案，原因有三：","无推理的检索：模型必须基于初始检索中选择的文本块回答问题，即使这些文本块不完整或无关。它无法决定在回答之前需要其他文档、其他部分或更多上下文，从而限制了模型推理的影响力。","文本块级限制：关键数据通常存在于复杂的多模态文档中。比如被问及“公司第三季度的有效税率是多少？”时，索引可能找到正确的文档，但无法打开文档、定位表格、阅读周围上下文或验证答案。","无迭代：许多问题需要多次检索才能得到正确答案。模型可能需要优化其搜索、检查有希望的文档、跟随引用、比较多个来源、跟踪已查看的信息，并在第一次结果不足时尝试新的途径。一站式RAG无法执行这些后续步骤。","没有代理式搜索（一站式检索）","仅使用1953年所有单独日历月份的报告值，美国国防及相关活动的支出总额（以百万名义美元计）是多少？","轨迹1 工具调用（仅搜索）","search(\"national defense expenditures monthly 1953\") → 10条命中：一些按月公告的零散信息（表3），每个覆盖财政年度的一部分，仅涵盖1953年的部分时间。","我找到了1953年1月至6月的数据。但我需要1953年7月至12月的月度数据才能计算答案。","轨迹 3 工具调用（2×搜索 → 阅读）","搜索(\"1953年国防开支每月\") → 每月公告（部分年份）","search(\"…1953年11月到1954年12月至今\") → 显示 treasury_bulletin_1954_02.pdf 第15页（表3，1953年的全部12个月）","read(treasury_bulletin_1954_02.pdf, p.15) → 提取完整的表3","Mistral Search Toolkit：https://docs.mistral.ai/en/studio/search-toolkit 提供用于在云端或本地摄取、嵌入和索引关键复杂数据的开放模块。代理式搜索基于该索引，为模型提供五种类似熟悉文件系统操作的工具：","search 使用现有索引在语料库中查找相关文档。","navigate 移动到页面、章节或其中的区域。","read 会检索该位置的内容。","grep 在打开的文档中查找匹配的模式。","模型不仅仅从初始的前k个结果回答问题，还可以检查找到的内容、优化搜索、打开相关文档、导航到特定部分并读取源材料后再回答。索引确定可能的来源；代理式搜索确定在这些来源内及之间需要检查的内容。","这些工具不需要微调或特定模型的训练。随着模型在推理和工具使用方面的能力提升，检索效果会在不改变基础设施的情况下提升。这是一个关键特性：检索质量随模型能力提升而提升，而不会被分块策略限制。","长文档。文件、合同、手册、技术规格和报告，其中答案可能出现在特定页面、特定表格、条款、图表或脚注中。","跨多个来源的问题。需要模型在得出答案之前，从多份文档中查找、比较或协调证据的研究。","必须验证的答案。财务数据、法律条款、监管参考和运营数据，其回答可以在一个稳定和特定的文档位置中被引用。","表格和结构化文档。财务报表、政府记录和扫描PDF，其中含义依赖于行、列、页面位置或周围上下文，而不仅仅是叙述文本。","直接查找。简短且干净的文档，答案可能出现在检索到的前几个分块中。","高量搜索。关键词或语义检索，需要返回相关段落，而不需要对其进行推理或浏览。","简单、可预测的问题。使用场景中答案的来源和位置事先已知，并且额外的检索步骤不太可能改善结果。","一次性RAG通常足以满足这些搜索需求。当问题需要模型超出初始结果并调查源材料时，添加Agentic搜索。配置良好的索引在两种情况下仍然是正确的基础。","我们使用开箱即用的Mistral搜索工具包堆栈对两个行业标准评估进行了Agentic搜索基准测试：默认分块、默认排序、无调优。这些结果是下限，而非上限，意味着你可以通过特定使用场景的调优进一步提升结果质量。","通过这些基准，我们使用Mistral搜索工具包测试了两个模型：Mistral Medium 3.5 (MM 3.5) 和 Z.ai GLM-5.2 (GLM-5.2)，展示了一个小模型(MM 3.5)和一个大型模型(GLM-5.2)的性能。","基准测试结果一致：代理循环提供了实质性的质量提升，导航工具在提高准确性的同时减少了无效的令牌、对话轮次和延迟。我们在一方和第三方模型中观察到相同的性能模式，这表明代理搜索（Agentic Search）与模型无关，并且随着新模型的出现，搜索质量应会提高。","FinanceBench（Islam 等, 2023）在368份美国证券交易委员会文件（10-K / 10-Q / 8-K）上测试财务问答，每份文件平均约147页，总计约53,900页：长篇且以表格为主的财务文件。答案由与人工标签校准的LLM评审进行评分。","仅搜索的代理循环是质量提升的最大杠杆。从一次性RAG（检索增强生成）到仅搜索循环，MM 3.5的准确率提升+47.3个百分点，GLM-5.2提升+52.6个百分点——对两种模型来说约为3倍的改进。因为模型可以迭代搜索，它们可以从初步结果的不足中恢复，改进查询，并将索引用作主动工具。","导航提升准确率。增加打开、导航、阅读和搜索（grep）功能再次提高准确率（MM 3.5 +8.7个百分点，GLM-5.2 +6.7个百分点）。这意味着在复杂文档中，针对性的深入搜索比反复的广泛搜索更有效。","更好的检索工具提高了令牌和性能效率。带导航的完整循环在使用比仅搜索循环更少的令牌的同时，回答更多问题正确（MM 3.5: 令牌使用-23.9%，GLM-5.2: -33.7%）。检索工具不是额外负担——它们用精准的导航替代了无效的搜索重试。","延迟在关键处下降。在FinanceBench中，添加导航检索工具提高了延迟表现：90百分位延迟从255秒降至154秒，平均延迟从108秒降至71秒。总体来看，我们观察到仅搜索循环会进行反复的广泛搜索，而导航帮助模型更快识别证据。","OfficeQA Pro是在历史美国财政部公报上的可验证数值基准：跨696份文档、约89,000页的扫描版、表格密集型政府财务PDF。我们报告了133个“Pro”子集问题的首次测试结果。","代理搜索和代理循环+导航在更严格、可验证的基准测试中表现出成功。OfficeQA Pro 包含数值答案、扫描 PDF 和复杂表格查找。即便在这种情况下，完整的代理循环也显著提升了准确性，相较于一次性 RAG，GLM-5.2 达到 51.9%（+45.6 个百分点），MM 3.5 提升了 +27.1 个百分点。","导航在提升质量的同时减少了浪费。使用完整循环（代理循环+导航）可以提高准确率最多 35.6%（MM 3.5 +7.5 个百分点；GLM-5.2 19.0% +8.3 个百分点），同时减少了 token 消耗。对话轮数下降最多 7.0%（MM 3.5，GLM-5.2 2.3%）。","基准越严格，检索循环的重要性越高。OfficeQA Pro 构建在扫描、表格密集文档中的数值答案上。一-shot RAG 几乎无法起步，而代理循环允许模型迭代搜索、检查证据并显著提高准确性。","工具堆栈对文档智能和搜索性能产生了显著影响。根据 Kimi 研究：https://www.kimi.ai/blog/kimi-k3，使用 Claude Code 束缚时，GLM-5.2 在 OfficeQA Pro 上得分 41.4%，而在 Mistral 束缚上得分为 51.9%——同一模型提升 +10.5 个百分点。","想了解更多关于代理搜索的信息，请查阅文档：https://docs.mistral.ai/studio/search/agentic-search。您可以在云端和本地部署中使用以下任一方式开始使用：","Mistral 搜索工具包：https://docs.mistral.ai/studio/search/search-toolkit。将 Agentic 搜索集成到您自己的代理、工作流程和客户部署中。","库：https://docs.mistral.ai/studio/libraries。在 Studio 和 Vibe 中开箱即用地使用代理搜索，无需自行构建检索系统。"]},"en":{"title":"Mistral Launches Agentic Search: Multi-Step Retrieval Improves AI System Accuracy in Complex Document Queries","summary":"Mistral releases Agentic Search, enabling multi-step retrieval loops through five tools—search, open, navigate, read, and grep—allowing the model to find, locate, and verify information across long documents and multiple sources. 🔗 Read the full article via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"Industry","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral Launches Agentic Search: Multi-Step Retrieval Improves AI System Accuracy in Complex Document Queries - Aioga AI News","description":"Mistral releases Agentic Search, enabling multi-step retrieval loops through five tools—search, open, navigate, read, and grep—allowing the model to find, locate, and verify inform...","url":"https://www.aioga.com/en/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:21.081Z"},"ja":{"title":"Mistralがエージェント検索を開始:多段階検索が複雑な文書を問い合わせるAIシステムの精度向上","summary":"MistralはAgentic Searchをリリースしました。これは検索、開く、ナビゲート、リード、grepの5つのツールからなる多段階検索ループを用いて、長文や複数の情報源にわたる情報の発見、位置特定、検証を可能にします。 🔗 原文記事はAIHOTより読むことができます。 https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"業界動向","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistralがエージェント検索を開始:多段階検索が複雑な文書を問い合わせるAIシステムの精度向上 - Aioga AIニュース","description":"MistralはAgentic Searchをリリースしました。これは検索、開く、ナビゲート、リード、grepの5つのツールからなる多段階検索ループを用いて、長文や複数の情報源にわたる情報の発見、位置特定、検証を可能にします。 🔗 原文記事はAIHOTより読むことができます。 https://aihot.virxact.com/items/cmt1pkwbj...","url":"https://www.aioga.com/ja/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:23.628Z"},"ko":{"title":"Mistral, Agentic Search 출시: 다단계 검색으로 AI 시스템의 복잡한 문서 조회 정확도 향상","summary":"Mistral이 Agentic Search를 발표했다. search, open, navigate, read, grep의 다섯 가지 도구를 통한 다단계 검색 루프를 통해 모델이 긴 문서와 여러 출처에서 정보를 찾아내고 위치를 확인하며 검증할 수 있다. 🔗 원문 읽기 via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"업계 동향","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral, Agentic Search 출시: 다단계 검색으로 AI 시스템의 복잡한 문서 조회 정확도 향상 - Aioga AI 뉴스","description":"Mistral이 Agentic Search를 발표했다. search, open, navigate, read, grep의 다섯 가지 도구를 통한 다단계 검색 루프를 통해 모델이 긴 문서와 여러 출처에서 정보를 찾아내고 위치를 확인하며 검증할 수 있다. 🔗 원문 읽기 via AIHOT · https://aihot.virxac...","url":"https://www.aioga.com/ko/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:30.306Z"},"es":{"title":"Mistral lanza Agentic Search: búsqueda de múltiples pasos mejora la precisión de consulta de documentos complejos en sistemas de IA","summary":"Mistral lanza Agentic Search, mediante un ciclo de búsqueda de múltiples pasos con cinco herramientas: search, open, navigate, read y grep, permitiendo que el modelo busque, localice y verifique información en documentos largos y de múltiples fuentes. 🔗 Leer el artículo original vía AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"Industria","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral lanza Agentic Search: búsqueda de múltiples pasos mejora la precisión de consulta de documentos complejos en sistemas de IA - Aioga Noticias de IA","description":"Mistral lanza Agentic Search, mediante un ciclo de búsqueda de múltiples pasos con cinco herramientas: search, open, navigate, read y grep, permitiendo que el modelo busque, locali...","url":"https://www.aioga.com/es/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:30.298Z"},"fr":{"title":"Mistral lance Agentic Search : une recherche en plusieurs étapes pour améliorer la précision des systèmes AI dans la consultation de documents complexes","summary":"Mistral publie Agentic Search, permettant à travers une boucle de recherche en plusieurs étapes utilisant cinq outils — search, open, navigate, read, grep — au modèle de rechercher, localiser et vérifier les informations dans de longs documents et sources multiples. 🔗 Lire l'article complet via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"Industrie","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral lance Agentic Search : une recherche en plusieurs étapes pour améliorer la précision des systèmes AI dans la consultation de documents complexes - Aioga Actualités IA","description":"Mistral publie Agentic Search, permettant à travers une boucle de recherche en plusieurs étapes utilisant cinq outils — search, open, navigate, read, grep — au modèle de rechercher...","url":"https://www.aioga.com/fr/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:38.011Z"},"de":{"title":"Mistral startet Agentic Search: Mehrstufige Suche verbessert die Genauigkeit von KI-Systemen, die komplexe Dokumente abfragen","summary":"Mistral veröffentlichte Agentic Search, das eine mehrstufige Suchschleife aus fünf Werkzeugen – Suchen, Öffnen, Navigieren, Lesen und Grep – verwendet, um dem Modell zu ermöglichen, Informationen über lange Dokumente und mehrere Quellen hinweg zu finden, zu lokalisieren und zu validieren. 🔗 Lesen Sie den Originalartikel über AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral startet Agentic Search: Mehrstufige Suche verbessert die Genauigkeit von KI-Systemen, die komplexe Dokumente abfragen - Aioga KI-News","description":"Mistral veröffentlichte Agentic Search, das eine mehrstufige Suchschleife aus fünf Werkzeugen – Suchen, Öffnen, Navigieren, Lesen und Grep – verwendet, um dem Modell zu ermöglichen...","url":"https://www.aioga.com/de/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:38.877Z"},"pt-BR":{"title":"Mistral lança a Busca Agente: busca em múltiplas etapas melhora a precisão dos sistemas de IA ao consultar documentos complexos","summary":"A Mistral lançou o Agentic Search, que utiliza um ciclo de busca em múltiplas etapas com cinco ferramentas — busca, abertura, navegação, leitura e grep — para permitir que o modelo encontre, localize e valide informações em documentos longos e múltiplas fontes. 🔗 Leia o artigo original via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral lança a Busca Agente: busca em múltiplas etapas melhora a precisão dos sistemas de IA ao consultar documentos complexos - Aioga Notícias de IA","description":"A Mistral lançou o Agentic Search, que utiliza um ciclo de busca em múltiplas etapas com cinco ferramentas — busca, abertura, navegação, leitura e grep — para permitir que o modelo...","url":"https://www.aioga.com/pt-BR/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:47.479Z"},"ru":{"title":"Mistral представила Agentic Search: многошаговый поиск повышает точность запросов к сложным документам ИИ-систем","summary":"Mistral выпустила Agentic Search, который с помощью цикла многошагового поиска с инструментами search, open, navigate, read и grep позволяет модели искать, находить и проверять информацию в длинных документах и из множества источников. 🔗 Читайте оригинал через AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral представила Agentic Search: многошаговый поиск повышает точность запросов к сложным документам ИИ-систем - Aioga Новости ИИ","description":"Mistral выпустила Agentic Search, который с помощью цикла многошагового поиска с инструментами search, open, navigate, read и grep позволяет модели искать, находить и проверять инф...","url":"https://www.aioga.com/ru/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:44.561Z"},"ar":{"title":"Mistral تطلق Agentic Search: تحسين دقة استعلامات الوثائق المعقدة لنظام الذكاء الاصطناعي من خلال البحث متعدد الخطوات","summary":"أصدرت Mistral Agentic Search، من خلال دورة استرجاع متعددة الخطوات باستخدام خمسة أدوات: البحث، الفتح، التنقل، القراءة، وgrep، لتمكين النموذج من البحث عن المعلومات وتحديد مواقعها والتحقق منها في مستندات طويلة ومن مصادر متعددة. 🔗 قراءة المقال الأصلي عبر AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral تطلق Agentic Search: تحسين دقة استعلامات الوثائق المعقدة لنظام الذكاء الاصطناعي من خلال البحث متعدد الخطوات - Aioga أخبار الذكاء الاصطناعي","description":"أصدرت Mistral Agentic Search، من خلال دورة استرجاع متعددة الخطوات باستخدام خمسة أدوات: البحث، الفتح، التنقل، القراءة، وgrep، لتمكين النموذج من البحث عن المعلومات وتحديد مواقعها وال...","url":"https://www.aioga.com/ar/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:53.702Z"},"hi":{"title":"Mistral ने पेश किया Agentic Search: बहु-चरण की खोज AI सिस्टम में जटिल दस्तावेज़ प्रश्नों की सटीकता बढ़ाती है","summary":"Mistral ने Agentic Search जारी किया, search, open, navigate, read, grep पांच उपकरणों के बहु-चरण खोज चक्र के माध्यम से, जिससे मॉडल लंबे दस्तावेज़ों और कई स्रोतों में जानकारी खोजने, स्थित करने और सत्यापित करने में सक्षम हो जाता है। 🔗 मूल लेख पढ़ें via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral ने पेश किया Agentic Search: बहु-चरण की खोज AI सिस्टम में जटिल दस्तावेज़ प्रश्नों की सटीकता बढ़ाती है - Aioga AI समाचार","description":"Mistral ने Agentic Search जारी किया, search, open, navigate, read, grep पांच उपकरणों के बहु-चरण खोज चक्र के माध्यम से, जिससे मॉडल लंबे दस्तावेज़ों और कई स्रोतों में जानकारी खोजने,...","url":"https://www.aioga.com/hi/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:22:54.892Z"},"it":{"title":"Mistral lancia Agentic Search: la ricerca a più fasi migliora la precisione delle query sui documenti complessi nei sistemi AI","summary":"Mistral rilascia Agentic Search, permettendo al modello di cercare, individuare e verificare informazioni in documenti lunghi e fonti multiple attraverso un ciclo di ricerca a più fasi con cinque strumenti: search, open, navigate, read e grep. 🔗 Leggi l'articolo originale via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral lancia Agentic Search: la ricerca a più fasi migliora la precisione delle query sui documenti complessi nei sistemi AI - Aioga Notizie IA","description":"Mistral rilascia Agentic Search, permettendo al modello di cercare, individuare e verificare informazioni in documenti lunghi e fonti multiple attraverso un ciclo di ricerca a più...","url":"https://www.aioga.com/it/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:01.217Z"},"nl":{"title":"Mistral lanceert Agentic Search: meerstapszoekfunctie verbetert de nauwkeurigheid van AI-systemen bij complexe documentopzoekingen","summary":"Mistral introduceert Agentic Search, waarmee het model informatie kan zoeken, lokaliseren en verifiëren in lange documenten en meerdere bronnen door een meerstapszoekcyclus met vijf tools: search, open, navigate, read en grep. 🔗 Lees het originele artikel via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral lanceert Agentic Search: meerstapszoekfunctie verbetert de nauwkeurigheid van AI-systemen bij complexe documentopzoekingen - Aioga AI-nieuws","description":"Mistral introduceert Agentic Search, waarmee het model informatie kan zoeken, lokaliseren en verifiëren in lange documenten en meerdere bronnen door een meerstapszoekcyclus met vij...","url":"https://www.aioga.com/nl/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:01.003Z"},"tr":{"title":"Mistral, Ajanik Aramayı başlattı: Çok adımlı arama, karmaşık belgeleri sorgulayan yapay zeka sistemlerinin doğruluğunu artırır","summary":"Mistral, modelin uzun belgeler ve birden fazla kaynak arasında bilgi bulmasını, bulmasını ve doğrulamasını sağlayan beş araçtan oluşan çok adımlı arama döngüsü kullanan Ajanic Search'i piyasaya sürdü. 🔗 Orijinal makaleyi AIHOT üzerinden okuyun · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral, Ajanik Aramayı başlattı: Çok adımlı arama, karmaşık belgeleri sorgulayan yapay zeka sistemlerinin doğruluğunu artırır - Aioga AI Haberleri","description":"Mistral, modelin uzun belgeler ve birden fazla kaynak arasında bilgi bulmasını, bulmasını ve doğrulamasını sağlayan beş araçtan oluşan çok adımlı arama döngüsü kullanan Ajanic Sear...","url":"https://www.aioga.com/tr/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:09.810Z"},"vi":{"title":"Mistral ra mắt Agentic Search: tìm kiếm nhiều bước nâng cao độ chính xác của hệ thống AI khi truy vấn tài liệu phức tạp","summary":"Mistral đã phát hành Agentic Search, sử dụng vòng lặp tìm kiếm nhiều bước gồm năm công cụ—tìm kiếm, mở, điều hướng, đọc và grep—để cho phép mô hình tìm, định vị và xác thực thông tin trên các tài liệu dài và nhiều nguồn. 🔗 Đọc bài viết gốc qua AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral ra mắt Agentic Search: tìm kiếm nhiều bước nâng cao độ chính xác của hệ thống AI khi truy vấn tài liệu phức tạp - Tin tức AI Aioga","description":"Mistral đã phát hành Agentic Search, sử dụng vòng lặp tìm kiếm nhiều bước gồm năm công cụ—tìm kiếm, mở, điều hướng, đọc và grep—để cho phép mô hình tìm, định vị và xác thực thông t...","url":"https://www.aioga.com/vi/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:09.832Z"},"id":{"title":"Mistral meluncurkan Agentic Search: Pencarian multi-langkah meningkatkan akurasi kueri dokumen kompleks pada sistem AI","summary":"Mistral merilis Agentic Search, melalui siklus pencarian multi-langkah dengan lima alat: search, open, navigate, read, grep, memungkinkan model untuk mencari, menempatkan, dan memverifikasi informasi dalam dokumen panjang dan dari berbagai sumber. 🔗 Baca artikel asli via AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral meluncurkan Agentic Search: Pencarian multi-langkah meningkatkan akurasi kueri dokumen kompleks pada sistem AI - Berita AI Aioga","description":"Mistral merilis Agentic Search, melalui siklus pencarian multi-langkah dengan lima alat: search, open, navigate, read, grep, memungkinkan model untuk mencari, menempatkan, dan memv...","url":"https://www.aioga.com/id/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:15.406Z"},"th":{"title":"Mistral เปิดตัว Agentic Search: การค้นหาหลายขั้นตอนช่วยเพิ่มความแม่นยําของระบบ AI ในการสอบถามเอกสารที่ซับซ้อน","summary":"Mistral ได้เปิดตัว Agentic Search ซึ่งใช้วงจรการค้นหาหลายขั้นตอนของเครื่องมือห้าชนิด ได้แก่ การค้นหา เปิด นําทาง อ่าน และ grep เพื่อให้โมเดลสามารถค้นหา ระบุตําแหน่ง และตรวจสอบข้อมูลจากเอกสารยาวและแหล่งข้อมูลหลายแหล่ง 🔗 อ่านบทความต้นฉบับผ่าน AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral เปิดตัว Agentic Search: การค้นหาหลายขั้นตอนช่วยเพิ่มความแม่นยําของระบบ AI ในการสอบถามเอกสารที่ซับซ้อน - ข่าว AI Aioga","description":"Mistral ได้เปิดตัว Agentic Search ซึ่งใช้วงจรการค้นหาหลายขั้นตอนของเครื่องมือห้าชนิด ได้แก่ การค้นหา เปิด นําทาง อ่าน และ grep เพื่อให้โมเดลสามารถค้นหา ระบุตําแหน่ง และตรวจสอบข้อมู...","url":"https://www.aioga.com/th/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:18.512Z"},"pl":{"title":"Mistral uruchamia wyszukiwanie agentyczne: wieloetapowe wyszukiwanie poprawia dokładność systemów AI w zapytaniach złożonych dokumentów","summary":"Mistral wypuścił Agentic Search, który wykorzystuje wieloetapową pętlę wyszukiwania składającą się z pięciu narzędzi — wyszukiwania, otwierania, nawigacji, odczytu i grepowania — aby umożliwić modelu znajdowanie, lokalizowanie i weryfikację informacji w długich dokumentach i wielu źródłach. 🔗 Przeczytaj oryginalny artykuł za pośrednictwem AIHOT · https://aihot.virxact.com/items/cmt1pkwbj04bxroovzkfca5c7","category":"行业动态","source":"Mistral AI：News（网页","aggregationSource":"Mistral AI：News（网页","pageTitle":"Mistral uruchamia wyszukiwanie agentyczne: wieloetapowe wyszukiwanie poprawia dokładność systemów AI w zapytaniach złożonych dokumentów - Aioga Wiadomości AI","description":"Mistral wypuścił Agentic Search, który wykorzystuje wieloetapową pętlę wyszukiwania składającą się z pięciu narzędzi — wyszukiwania, otwierania, nawigacji, odczytu i grepowania — a...","url":"https://www.aioga.com/pl/news/cmt1pkwbj04bxroovzkfca5c7/","contentTranslated":true,"sourceHash":"54e6108ecb6e765b","translatedAt":"2026-08-20T16:23:27.158Z"}},"evidenceTier":"verified-news","reviewStatus":"editorial-selected","indexable":true,"editorialCover":""}}