11-19 更新 实时
首页/ AI 大模型/ 正文

SID AI 发布 SID-2 智能体式搜索模型

AI 大模型 来源:MarkTechPost 2026-11-18 15:10 0 阅读 机器翻译

SID AI 发布了 SID-2,这是其第二代智能体式搜索模型,同时发布的还有 SID-2 Swarm。SID-2 是 SID AI 推出的智能体式搜索模型,它会搜索你的数据、读取结果并返回排序后的文档 ID。它用一个自行决定下一步搜什么的模型,取代了固定的 RAG 流水线。

SID-2 是一个小型的搜索专用模型,而非通用聊天模型。静态流水线按固定顺序依次执行改写、嵌入、搜索和重排。而 SID-2 则是搜索、读取结果、调整后再搜索。

开发者针对自己的后端实现搜索工具。text_search 工具负责关键词搜索,search 工具负责向量搜索。模型在每一轮结束时调用 report_helpful_ids,返回一个排序列表。由于 SID-2 只能看到你的工具返回的内容,因此无需重建索引或迁移数据。

SID-2 Swarm 会运行多个智能体,它们相互发送消息,并把各自的上下文分叉给新的智能体。每个智能体负责探索语料库中一个单独的切片。SID 认为,搜索的收益更多来自阅读更多文档,而不是对每篇文档做更深入的推理。

SID-2 提供低、中、高三档推理强度。每一档都作为独立的专家进行训练,随后通过多教师同策略(on-policy)蒸馏合并。Metric Steering 让用户可以以 NDCG、F1 或 Recall@k 为目标,其中 k 从 1 到 50。在 SID 的 506 题验证集上,高推理强度的 SID-2 在每项指标上都比基础模型高出 0.14 至 0.22。

训练采用两阶段强化学习。第 1 阶段只优化检索质量。第 2 阶段加入一个乘法形式的延迟惩罚,且仅作用于正向奖励。一个提议者-求解者自博弈循环针对冻结的 SID-1 求解器生成高难度查询。SID 报告称,相比 SLIME 的默认设置,训练吞吐提升了 6 倍。

SID 构建了一个包含 506 个问题、覆盖 19 个语料库的基准测试。这些语料库包括 HotpotQA、Enron 邮件、SEC 文件、PubMed 和 BrowseComp-Plus。下文所有数字均为该报告表 2 至表 4 中的自报数据。

Swarm 在 19 个语料库中的 11 个上领先。在 WixQA 上,单个高投入智能体以约低 232 倍的成本达到与 Astra 相当的水平。在最低投入下,SID-2 以少 13 倍的 FLOPs 胜过 4 个并行运行的 SID-1。

NDCG、时间和成本各行由 SID 在其自有基准测试上测量。检索后端耗时不计入其中。

SID-2 可通过 SID 的 API 以及 Baseten 使用。已提供 Python(sid-sdk)、TypeScript(@sid-ai/sid-sdk)和 Go(github.com/sidhq/sid-go)的 SDK。一个 SageMaker 辅助工具可支持 AWS 部署。参考实现采用 OpenAI 客户端格式。

SID-2 的价格为每百万输入 Token 0.25 美元、每百万输出 Token 1.00 美元。Swarm 为 0.50 美元和 2.00 美元。缓存输入价格为输入价格的 10%。

查看 SID-2 技术报告、文档和模型页面。你可以通过我们的 GPULeaderboard.com Chrome 扩展在此查看部署选项和规格。

需要与我们合作推广你的 GitHub 仓库、Hugging Face 页面、产品发布或网络研讨会等?欢迎联系我们。

题为《SID AI 发布具备 0.720 NDCG 的 SID-2 智能体式搜索模型,含 Swarm 模式》的文章首发于 MarkTechPost。

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(SID AI Releases SID-2 Agentic Search Model With 0.720 NDCG Swarm Mode)

SID AI has released SID-2, its second-generation agentic search model, alongside SID-2 Swarm. SID-2 is an agentic search model from SID AI that searches your data, reads results and returns ranked document IDs. It replaces fixed RAG pipelines with a model that decides what to search next.

TL;DR

  • Size: Context window is 262k tokens per the report (256k on sid.ai/models).
  • Runs on: Hosted API at platform.sid.ai, plus Baseten and SageMaker reference paths. SID’s latency figures assume 1 B200 GPU at 8 concurrent queries.
  • Performance: A single agent matches Claude Opus 5.5 on retrieval quality. Swarms beat every model SID tested.
  • Best: SID-2 Swarm (12 agents) scores 0.720 NDCG, above GPT-6 Astra at 0.704.
  • Worst: SID-2 Low scores 0.556 NDCG, 0.148 below GPT-6 Astra. Astra still leads on California law, SEC filings and FiQA.
  • Bottom line: Best: frontier-level retrieval at a fraction of the latency. Worst: no disclosed size, weights or independent evals yet.

What is SID-2 and how does it work?

SID-2 is a small search specialist, not a general chat model. A static pipeline runs rewrite, embed, search and rerank in a fixed order. SID-2 instead searches, reads results, adapts and searches again.

Developers implement the search tools against their own backend. A text_search tool covers keyword search and a search tool covers vector search. The model ends each episode by calling report_helpful_ids with a ranked list. No reindexing or data movement is required, since SID-2 only sees what your tools return.

SID-2 Swarm runs several agents that message each other and fork their context into new agents. Each agent explores a separate slice of the corpus. SID argues search gains more from reading more documents than from reasoning harder about each one.

What are reasoning effort and Metric Steering?

SID-2 offers low, medium and high reasoning effort. Each level was trained as a separate expert, then merged with multi-teacher on-policy distillation. Metric Steering lets users target NDCG, F1 or Recall@k, with k from 1 to 50. On SID’s 506-question validation set, SID-2 at high effort scores 0.14 to 0.22 higher per metric than the base model.

How was SID-2 trained?

Training used 2-stage reinforcement learning. Stage 1 optimizes retrieval quality only. Stage 2 adds a multiplicative latency penalty that applies only to positive rewards. A proposer-solver self-play loop generated hard queries against a frozen SID-1 solver. SID reports a 6x training-throughput gain over SLIME’s default setup.

How does SID-2 perform on benchmarks?

SID built a 506-question benchmark across 19 corpora. These include HotpotQA, Enron emails, SEC filings, PubMed and BrowseComp-Plus. All numbers below are self-reported from Tables 2 to 4 of the report.

  • SID-2 Medium: 0.585 NDCG in 3.11 s, versus Claude Opus 5.5 (medium) at 0.577 in 25.1 s.
  • SID-2 Swarm, 6 agents: 0.701 NDCG in 17.7 s, versus GPT-6 Astra (max) at 0.704 in 135 s.
  • SID-2 Swarm, 12 agents: 0.720 NDCG, the highest score in the table.
  • Baselines: BM25 scores 0.260 and vector search plus reranker scores 0.287.

Swarm leads on 11 of 19 corpora. On WixQA, a single high-effort agent matches Astra at about 232x lower cost. At its lowest effort, SID-2 beats 4 parallel SID-1 runs with 13x fewer FLOPs.

SID-2 vs GPT-6 Astra vs Claude Opus 5.5

FeatureSID-2 (medium)SID-2 Swarm (6 agents)GPT-6 Astra (max)Claude Opus 5.5 (medium)
TypeAgentic search modelMulti-agent searchGeneral LLMGeneral LLM
ParametersNot disclosedNot disclosedNot disclosedNot disclosed
Context262k (report) / 256k (site)Effectively infinite (per SID)Not disclosed1M
NDCG, SID benchmark0.5850.7010.7040.577
Inference time per query3.11 s17.7 s135 s25.1 s
Cost per 1,000 queries$6.54$263$2,141$131
API price (input / output per 1M)$0.25 / $1.00$0.50 / $2.00$10 / $50$4 / $20

NDCG, time and cost rows are measured by SID on its own benchmark. Retrieval backend time is excluded.

How can developers deploy SID-2?

SID-2 is available through SID’s API and on Baseten. SDKs exist for Python (sid-sdk), TypeScript (@sid-ai/sid-sdk) and Go (github.com/sidhq/sid-go). A SageMaker helper covers AWS deployments. The reference implementation uses the OpenAI client format.

SID-2 costs $0.25 per million input tokens and $1.00 per million output tokens. Swarm costs $0.50 and $2.00. Cached input is 10% of the input price.

Key Takeaways

  • SID-2 is an RL-trained agentic search model that returns ranked document IDs.
  • SID-2 Medium matches Opus 5.5 retrieval at 0.585 NDCG in 3.11 s.
  • 12-agent SID-2 Swarm leads SID’s benchmark at 0.720 NDCG.
  • Pricing starts at $0.25 per million input tokens, with no reindexing.
  • All results are self-reported; size and weights are undisclosed.


Check out the SID-2 technical report, the docs and the model page. You can check the deployment options and specs here through our GPULeaderboard.com chrome extension.

Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.

Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us

The post SID AI Releases SID-2 Agentic Search Model With 0.720 NDCG Swarm Mode appeared first on MarkTechPost.

正文保留英文原文,标题与摘要由机器翻译。

本文由采集器自动同步

原文发布于 MarkTechPost(SID AI Releases SID-2 Agentic Search Model With 0.720 NDCG Swarm Mode),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 ↗