Metaso (秘塔AI搜索, branded "META SOTA" in its logo, at metaso.cn) is an AI search engine operated by Shanghai Metaso Network Technology Co., Ltd. It launched in February 2024 (BAAI Hub interview, ScienceNet report). Instead of a list of links, it answers a question with a structured summary whose claims carry inline citations to a visible source list. Its verifiable distinctions among AI search products: it is Chinese-first, claims to be ad-free, requires no account for basic searches, and offers a Deep Research mode free of charge. Note that the site has no English interface — the product is documented here because it is one of the most widely used AI search tools in the Chinese-speaking world, but readers will need to navigate Simplified Chinese UI.

At a Glance

  • URL: https://metaso.cn/
  • Type: AI search engine (web app plus mobile and Windows clients)
  • Cost: Basic search is free and ad-free; paid membership and a metered search API also exist
  • Registration: Not required for basic search; Deep Research, history and library features require sign-in (WeChat scan, phone SMS code, or account password)
  • Interface language: Simplified Chinese only; brand mark reads META SOTA

(Verified by direct visit on 2026-08-31.)

Background

Shanghai Metaso Network Technology was founded in April 2018 and is headquartered in Shanghai's "Model Speed Space" AI hub. The company started with legal-tech AI tools, then expanded into general productivity products: the Xiezuocat writing assistant, Metaso Translate, and MetaLaw, a legal case-search engine. Its self-developed large language model MetaLLM passed China's generative-model filing in 2023 (official About page).

Founder and CEO Min Kerui previously served as chief scientist at Cheetah Mobile and CTO of Boson Data. In August 2024 the company raised over RMB 100 million in a round led by Ant Group with Lightspeed China participating (Economic Observer). The company claims tens of millions of users across its product line, per its own About page.

Search Capabilities in Practice

Metaso homepage with three search modes: concise, in-depth, and Deep Research

The homepage is a single search box. Each query runs in one of three modes (tested 2026-08-31):

  • Concise (简洁): a short, quick answer;
  • In-depth (深入): the default, a fuller explanation;
  • Deep Research (深度研究): multi-round retrieval and reasoning that produces a full research report. It has been free to the public since July 2025 (QbitAI report), but currently requires signing in first.

Search scopes can be switched among web, documents (文库), academic papers, images, videos, and podcasts. On the results page, the structured answer sits on the left while the full citation list stays visible on the right for cross-checking:

A web search results page: structured answer on the left, citation list on the right

Academic search mode: the answer is organized by topic and sources are paper records

Answers can also be rendered as outlines and mind maps, and further converted into slide decks or interactive web reports. The homepage additionally surfaces learning-oriented entries ("学点啥") and video generation. For developers, Metaso opened a search API in July 2025 priced at RMB 0.03 per query, backed by a self-built multilingual index of tens of billions of pages (OSChina report).

Mobile clients exist for iOS and Android (Android package com.metaso, Tencent MyApp listing), and a Windows version is available in the Microsoft Store.

Pricing and Accounts

Basic search is free, ad-free, and usable without an account. Metaso uses a daily search-credit system: as reported in July 2025, free accounts get roughly 100 credits per day and one Deep Research run consumes about 20–30 credits; a user-reported annual membership costs about RMB 179 and raises the daily quota to 500 credits (hands-on report via Tencent News). There is no public pricing page on the official site, so these third-party figures may have changed as of 2026-08-31 — treat the in-app display after sign-in as authoritative.

On privacy, the official About page states that traffic is encrypted over HTTPS/TLS and user files are stored encrypted.

Good Use Cases

  • Getting a quick, source-cited overview of a topic discussed mainly in Chinese;
  • Literature screening: academic mode cites paper records (after CNKI sent an infringement notice in August 2024, Metaso said it stopped indexing CNKI metadata and switched to other Chinese and English knowledge bases — ScienceNet — so coverage differs from international indexes);
  • Users in mainland China who need an AI search engine that works without network workarounds;
  • Developers who need a Chinese web-search API.

Limitations

  • AI answers can still be wrong: every results page carries a standing "content is AI-generated, please verify" disclaimer; key claims should be checked against the cited sources.
  • Some features require sign-in: as of 2026-08-31, starting Deep Research triggers a WeChat-scan login wall; history and library features also depend on an account.
  • Chinese-first product: the UI is Simplified Chinese only, and coverage of English queries and English-language sources is weaker than international alternatives.
  • Opaque pricing: no official public pricing page; quota and membership figures come from third-party reports and may change at any time.
  • "Tens of millions of users" is the company's own claim, without per-product or as-of-date breakdown.

Compared with Perplexity

The closest counterpart, already covered on this site, is Perplexity: launched in December 2022, English-first and oriented toward international sources, with advanced features gated behind a US$20/month Pro subscription. Metaso targets Chinese-language scenarios with a more generous free tier and direct accessibility in mainland China. Technically, Perplexity initially relied on Google and Bing search APIs with retrieval-augmented generation (TMTPost analysis), while Metaso says it runs on a self-built index and its in-house MetaLLM model. A practical split: use Metaso for Chinese queries, Perplexity for English and international research.

References