UI-TARS-desktop

bytedanceApache-2.0活跃维护

字节开源的多模态 AI Agent 桌面栈,把前沿模型和 Agent 基础设施连起来,让 Agent 能像人一样操作 GUI、浏览器和电脑完成任务。

The Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra

v0.3.1发布于 2026-10-08 21:14 UTC39,586star +21 / 7 天4,030fork同步于 2026-11-09 11:19 UTC

怎么上手

从 README 代码块原样提取,可能是其中一种安装方式
npx @agent-tars/cli@latest
npm install @agent-tars/cli@latest -g
硬件要求

README 没有说明硬件要求。

下载:源码 zip·GitHub Releases(安装包)本站不镜像安装包

项目说明

这个项目的说明还在生成中,先参考下面「最近发版」和 README。

最近发版

5 个版本v0.3.1
  1. v0.3.1最新正式版2026-10-08 · 2026-10-08 21:14 UTC

    This release closes several security issues in the Agent TARS CLI server (@agent-tars/cli, @tarko/agent-server). It changes how the server can be reached, so please read Upgrading from 0.3.0 before you update.

    Upgrading from 0.3.0

    If you run the CLI on your own machine and open it at http://localhost, nothing changes.

    Otherwise:

    • The server only listens on 127.0.0.1 by default. It used to listen on all network interfaces. To reach it from another machine or from outside a container, pass --host, e.g. --host 0.0.0.0.
    • A non-loopback --host requires an access token. Set one with --auth-token or TARKO_AUTH_TOKEN (at least 16 characters). If you set none, a random token is generated on every start and printed with a ready-to-open ?token= link, so old links stop working after a restart. Set a fixed token if you need stable links.
    • Domains, reverse proxies and port forwarding need TARKO_ALLOWED_HOSTS. Requests whose Host header is not a loopback name, an IP address or the configured --host are rejected with 403 Invalid Host header. Add your hostname, comma-separated: TARKO_ALLOWED_HOSTS=agent.example.com.
    • A Web UI served from a non-localhost origin needs TARKO_ALLOWED_ORIGINS. Otherwise creating a session fails with 403 Origin not allowed. Use the full origin: TARKO_ALLOWED_ORIGINS=https://agent.example.com.
    • agentOptions from requests and URLs only accepts agentMode. This applies to /creating?agentOptions=..., welcome cards and POST /api/v1/sessions/create. Any other key (mcpServers, model, aioSandbox, ...) is rejected with 400. Configure those on the server instead.
    • Direct API calls need a CSRF token. Fetch one from GET /api/v1/csrf-token and send it as the x-csrf-token header on POST requests. When the server requires an access token, also send Authorization: Bearer <token>.

    Example for a remote server behind a domain:

    TARKO_AUTH_TOKEN=<a long random string> \
    TARKO_ALLOWED_HOSTS=agent.example.com \
    TARKO_ALLOWED_ORIGINS=https://agent.example.com \
    agent-tars --host 0.0.0.0 --port 8888
    

    See the CLI docs for details.

    Security

    • Block request-body injection into agent constructor options; the server now binds to 127.0.0.1 by default (#1939)
    • Require an access token once the server leaves loopback (#2026)
    • Validate the Host header to prevent DNS rebinding (#1975)
    • Add CSRF protection, a CORS allowlist and security headers (#1853)
    • Stop HTML previews from escaping their iframe sandbox (#1938)

    Bug Fixes

    • Answer a disallowed origin with 403 instead of 500 (#2044)
    • Repair unterminated strings in truncated tool call JSON (#1836)
    • Handle unknown model providers by defaulting to OpenAI-compatible (#1823)
    • Support streaming usage on LLM responses (#1737)

    Features

    • Add embed-frame support for nav items (#1761)
    • Optimize the session status action design (#1766)
    • Set the remote browser viewport to the default viewport (#1748)
    • Add @gui-agent/cli, a CLI for GUI Agent automation (#1741)

    Documentation

    • Document remote access settings for the agent server (#2043)

    Full Changelog: https://github.com/bytedance/UI-TARS-desktop/compare/v0.3.0...v0.3.1

  2. v0.3.02025-11-04 · 2025-11-04 16:10 UTC
    查看更新说明

    What's Changed

    New Features 🎉

    • feat(gui-agent): add example for 2.0 version (#1732) by @ZhaoHeh in 39ee4a8
    • feat(gui-agent): add image detail calculator for enhanced screenshot processing (#1724) by @ZhaoHeh in 71e6e1a
    • feat(aio-operator): remove focus from the address bar after navigation (#1723) by @ZhaoHeh in 9fb722b
    • feat(o-gui-agent): add image's detail parameter into message (#1722) by @ZhaoHeh in 53bb572
    • feat(tarko-agent-ui): add copy link option to share conversation modal (#1706) by @ULIVZ in 89b0021
    • feat(tarko): add event stream viewer (#1374) by @ULIVZ in 41bf5b9
    • feat(game-mode): improve game agent mode error handling and user experience (#1673) by @ZhaoHeh in 41771b5
    • feat(tarko-agent-ui): improve raw mode renderer design (#1677) by @ULIVZ in 7aca7b3
    • feat(agent-tars): add welcome cards (#1685) by @ULIVZ in 20aefb1
    • feat(tarko-server-next): optimize parts of controller by avoiding go … (#1684) by @cjraft in 3e4dbb1
    • feat(omni-agent): fix img compressor dependency issues (#1683) by @cjraft in 135cf99
    • feat: TARS platform enhancements and compatibility fixes (#1680) by @cjraft in c350778
    • feat(tarko-agent-ui): add clickable url links in terminal renderer json output (#1655) by @ULIVZ in 3d18050
    • feat(tarko-agent-ui): replace plus icon with lightning bolt for agent mode selector (#1674) by @ULIVZ in 4375ec1
    • feat(omni-agent): zoom out game welcome card (#1672) by @ZhaoHeh in 9dc0d05
    • feat(o-gui-agent): set waiting time as 5s when initailize screenshot (#1671) by @ZhaoHeh in 3b6ca2b
    • feat(omni-agent): add functions prompt for game mode (#1668) by @ZhaoHeh in eab67bf
    • feat(omni-agent): optimize agent game mode UI (#1665) by @ZhaoHeh in db0aade
    • feat(tarko-agent-next): sync code from agent-server (#1663) by @cjraft in 2602b10
    • feat(tarko-agent-server, agent-ui): return initialization events from session create (#1662) by @ULIVZ in 008124a
    • feat(o-gui-agent): insert preset messages before real query from user (#1661) by @ZhaoHeh in 43ba4cf
    • feat(omni-agent): add default runtimeSettings (#1657) by @cjraft in 464ee9b
    • feat(o-gui-agent): support agent mode and game operator with enhanced error handling (#1656) by @ZhaoHeh in 4539b80
    • feat(tarko-agent-server, agent-ui): refine arch of agent options and settings from home to session (#1645) by @ULIVZ in 3a27504
    • feat(omni-agent): agentMode structure update; add game mode support (#1649) by @cjraft in f3ec7d4
    • feat(omni-agent): compress images for gui screenshot (#1647) by @cjraft in b8e3d68
    • feat(tarko-agent-ui): welcome cards (#1643) by @ULIVZ in 37babcc
    • feat(omni-agent): update agent mode config structure (#1642) by @cjraft in c3e8513
    • feat(tarko-agent): enhance runtime settings with enum labels and UI placement control (#1638) by @ULIVZ in 70571c2
    • feat(tarko-agent-ui): replaced generic tool renderer with terminal-style renderer (#1594) by @ULIVZ in 6367f5a
    • feat(tarko): add runtime settings api in server-next (#1634) by @cjraft in 8d6b08e
    • feat(tarko-agent-agent, agent-ui): add webui.base support (#1623) by @ULIVZ in dce0e7e
    • feat(tarko-agent-ui): add webui.layout.enableHome config (#1629) by @ULIVZ in d6673d5
    • feat(tarko-agent-ui): add webui.layout.enableSidebar config (#1624) by @ULIVZ in 530c435
    • feat(agent-tars): disable contextual selector by default (#1621) by @ULIVZ in 86f3da0
    • feat(tarko-agent): runtime settings (#1561) by @ULIVZ in a59d33d
    • feat(tarko-agent-cli): add --model.displayName option support (#1605) by @ULIVZ in dadfe4e
    • feat(tarko-agent-ui): mobile ui bottom sheet (#1563) by @ULIVZ in 0a1b7cf
    • feat(agent-tars): support aio sandbox environment (#1573) by @ULIVZ in 2e38422
    • feat(tarko-agent): remove engine-selector from agent kernel (#1576) by @ULIVZ in 76da098
    • feat(agent-server): support remote static html and add share for server-next (#1558) by @cjraft in 87dd3b9
    • feat(tarko-agent-cli): support unknown options passthrough (#1574) by @ULIVZ in f1f2ccd
    • feat(tarko-agent-ui): new layout de
  3. v0.3.0-beta.122025-09-19 · 2025-09-19 02:13 UTC
    查看更新说明

    What's Changed

    New Features 🎉

    • feat(tarko-agent-ui): new layout design (#1553) by @ULIVZ in 21d59fb
    • feat(tarko-agent-server-next): support replacing server global logger (#1554) by @cjraft in 4228662
    • feat(tarko-agent-server-next): add initial events for agent when change model (#1552) by @cjraft in 55ae6d7
    • feat(tarko-agent): restore events on agent session restore (#1548) by @ULIVZ in 1282b70
    • feat: add @tarko/agent-server-next (#1551) by @cjraft in 043a901
    • feat(tarko-agent-ui): disable model selector during agent run (#1520) by @ULIVZ in 751be28
    • feat(tarko-agent-ui): sunsetting plan functionality (#1521) by @ULIVZ in 5d875e2
    • feat(tarko-agent-server): add server.models support (#1488) by @ULIVZ in 0f4fd41
    • feat(model-provider): add automatic claude headers support (#1314) by @ULIVZ in 160b758
    • feat(tarko-mcp-agent): remove deprecated mcpClientVersion support (#1510) by @ULIVZ in 4f69ba2
    • feat(tarko-mcp-agent): adjust defaultConnectionTimeout from 180 to 60 seconds (#1505) by @ULIVZ in 4db1baf
    • feat(tarko): sunsetting model.providers from agent kernel (#1483) by @ULIVZ in 23002fd
    • feat(tarko-agent-ui): support line breaks in LinkReaderRenderer (#1482) by @ULIVZ in f72e813
    • feat(tarko-agent-ui): enhance agent ui config system (#1464) by @cjraft in d2f3633
    • feat(tarko-agent-ui): display colored terminal output (#1476) by @ULIVZ in c9d935b
    • feat(agent-tars): emphasize python3 usage in system prompt (#1469) by @ULIVZ in b457234
    • feat(tarko-mcp-agent): defaultConnectionTimeout option (#1462) by @ULIVZ in f49b2d7

    Bug Fixes 🐛

    • fix(tarko-agent-ui): support all SSE line separators in streaming (#1568) by @ULIVZ in bdceb7c
    • fix(tarko-agent-ui): tailwind cannot transpile styles from @tarko/ui (#1545) by @ULIVZ in a1f2c53
    • fix(tarko-agent-ui): config issues (#1541) by @cjraft in 34bc5b0
    • fix(tarko-agent-ui): navbar overlap on small screens (#1519) by @ULIVZ in 18e1de6
    • fix(model-provider): native providers regression issue (#1503) by @ULIVZ in a450de0
    • fix(tarko-agent): prevent max-iter warning in test by @ulivz in fdbf085
    • fix(tarko-agent): should track elapsed time for tool execution errors (#1474) by @ULIVZ in 6e7dd1f
    • fix(tarko-agent): wrong agent_run_end status (#1473) by @ULIVZ in c3a110b
    • fix(tarko-agent-ui): prevent undefined extension in code editor header (#1472) by @ULIVZ in 9b99a7d
    • fix(tarko): preserve logo param in session routing (#1471) by @ULIVZ in 9bff541

    Documentation 📚

    • docs(tarko-agent): init documentation (#1502) by @ULIVZ in b7af1e8
    • docs: remove duplicate header (#1559) by @ULIVZ in 52c0616
    • docs: correct typo in mcp documentation (#1516) by @ULIVZ in 435ea77
    • docs: refine useCursor implementation (#1501) by @ULIVZ in 760072b
    • docs: upgrade rspress to v2.0.0-beta.32 (#1495) by @ULIVZ in 0bac513
    • docs(agent-tars): agent api documentation (#1459) by @ULIVZ in 04cc246
    • docs: add @tarko/mcp-agent (#1458) by @ULIVZ in df34b24

    Other Changes

    • refactor(gui-agent): refactor gui agent sdk: agent, action-parser and operator (#1550) by @ZhaoHeh in ad242a5
    • refactor(tarko-agent-ui): refine ui infrastructure (#1546) by @ULIVZ in c5f4b6e
    • refactor(tarko): sunsetting web socket (#1543) by @ULIVZ in 22623f9
    • refactor(tarko-agent-ui): remove final_answer event handling (#1542) by @ULIVZ in 4e6b89e
    • refactor(tarko-agent-ui): make core low-level components stable (#1540) by @ULIVZ in 16c8660
    • refactor(tarko-agent-ui): remove @headlessui/react (#1539) by @ULIVZ in bb6f014
    • refactor(tarko-agent-ui): migrate stable editor atomics to @tarko/ui (#1537) by @ULIVZ in e8d7659
    • refactor(tarko-agent-ui): remove redundant motion animations (#1530) by @ULIVZ in 404367f
    • refactor(tarko-agent-ui): remove parrot comments (#1528) by @ULIVZ in 4873fb8
    • refactor(tarko-agent-ui): deduplicate muiTheme definitions (#1523) by @ULIVZ in c603b6e
    • refactor(tarko-agent-ui): remove redundant code and si
  4. v0.3.0-beta.112025-09-09 · 2025-09-09 14:44 UTC
    查看更新说明

    What's Changed

    From v0.3.0-beta.11, we have mainly introduced the Real-time Thinking Duration Protocol, introduced the MongoDB Provider to Agent Server, supported Tarko Agent UI Builder and CLI, and optimized a lot of Web UI experience.

    Here is a simple demo of Thinking Duration:

    https://github.com/user-attachments/assets/ee2e2883-4bf5-4558-a1a6-eff34916fb68

    New Features 🎉

    • feat(tarko): implement MongoDB provider for agent server (#1450) by @cjraft in b69aa5a
    • feat(tarko): agui cli for agent ui builder, see @tarko/agent-ui-cli (#1446) by @ulivz in 7bb9184
    • feat(o-agent): update sandbox sdk and gui-agent operator (#1437) by @cjraft in 8e2d7bb
    • feat(mcp-servers): support mcp offical registry (#1447) by @ycjcl868 in 5d773cf
    • feat(tarko): add navbar logo display options (#1443) by @ulivz in 4b1ed1f
    • feat(tarko): init @tarko/agent-ui-builder (#1436) by @ulivz in a99ac0c
    • feat(tarko): move workspace navItems from header to navbar (#1441) by @ulivz in 73fa2dc
    • feat(tarko): add tabbed file viewer for read_multiple_files tool (#1438) by @ulivz in 88f3568
    • feat(gui-agent): improve page visibility detection in AIOBrowser (#1431) by @ZhaoHeh in 230853e
    • feat(tarko-agent): thinking duration protocol and modernize thinking ui (#1423) by @ulivz in 094d40e
    • feat(tarko): refine collected files (#1422) by @ulivz in 95b1bfb
    • feat(tarko): add guiAgent.renderBrowserShell option (#1421) by @ulivz in 5a9d8e4

    Bug Fixes 🐛

    • fix(tarko): fetch actual remote config instead of local file (#1449) by @ryanroe in 083f842
    • fix(tarko): external @tarko/agent-ui-builder in agent-cli build (#1445) by @ulivz in fe579ae
    • fix(tarko): improve markdown inline code wrapping (#1439) by @ulivz in df9f553
    • fix(tarko): resolve react key spread warning and hooks render issue (#1435) by @ulivz in f3f4bf6
    • fix(tarko): make thinking toggle default expanded without initial animation (#1432) by @ulivz in ce0947d
    • fix(tarko): prevent frequent api/v1/models calls by memoizing callbacks (#1378) by @ulivz in e07ec41
    • fix(tarko): improve scroll-to-bottom indicator edge case handling (#1429) by @ulivz in 50eb9f2
    • fix(tarko): prevent duplicate session loading in SessionRouter (#1427) by @ulivz in f96d4ff
    • fix(tarko-agent): improve JSON parsing in PromptEngineeringToolCallEngine (close: #1360) (#1361) by @ulivz in b2d5817

    Other Changes

    • refactor(tarko): rename SessionItemInfo to SessionInfo (#1440) by @ulivz in d1b4d97
    • refactor(tarko): rename agent-web-ui to agent-ui (#1434) by @ulivz in 9a4e8f5
    • refactor(tarko): simplify code editor components (#1425) by @ulivz in 8b46f6f
    • chore(tars-stack): release 0.3.0-beta.11 by @ulivz in be3cfab
    • chore(all): fix changelog generation (#1420) by @ulivz in e53360b
    • ci: remove tag prefix from release scripts (#1451) by @ulivz in 12ebdba
    • ci(ptk): github release (#1428) by @ulivz in cbe3894

    Full Changelog: @[email protected]

  5. @[email protected]v0.3.0-beta.102025-09-07 · 2025-09-07 19:32 UTC
    查看更新说明

    What's Changed

    News ✨

    We are thrilled to announce that this version officially introduces support for UI-TARS-2! 🎉

    This milestone, together with v0.3.0-beta.9, represents a new beginning for the TARS Stack. 🚀

    This release wouldn't have been possible without the dedication and hard work of our incredible Core Contributors:
    @cjraft, @ZhaoHeh, @ycjcl868, @skychx, @ulivz, and @helio9cn. 🙌

    The journey has just begun—stay tuned for more exciting updates! 💡✨

    Demo

    https://github.com/user-attachments/assets/524a9f0f-b669-4533-9a59-09d934ef4340

    https://github.com/user-attachments/assets/c1b44eb0-3e21-42d8-b32c-a8b080bf366e

    New Features 🎉

    • feat(tarko): limit welcome prompts to 3 with shuffle (#1416) by @ulivz in c6d6791
    • feat(tarko): refine all empty state (#1408) by @ulivz in 18dc008
    • feat(tarko): add user message auto-scroll in normal mode (#1412) by @ulivz in 2c7f55d
    • feat(tarko): enhance slug generation with multilingual support (#1410) by @ulivz in 915c7c5
    • feat(tarko): auto-scroll for replay (#1407) by @ulivz in da22a39
    • feat(tarko): improve ChatInput UX with conditional help text and home variant (#1406) by @ulivz in 8c38bfc
    • feat(tarko): refine thinking animation (#1404) by @ulivz in bae4951
    • feat(tarko): refine scroll-to-bottom indicator (#1402) by @ulivz in 3a7d239
    • feat(tarko): defaults background to white for html renderer (#1397) by @ulivz in c583e7e

    Bug Fixes 🐛

    • fix(tarko): prevent auto-scroll on refresh for historical user messages (#1415) by @ulivz in 62df723
    • fix(tarko): improve scroll-to-bottom indicator detection (#1411) by @ulivz in 556e3a0
    • fix(tarko): improve session UI state management (#1409) by @ulivz in 0391c11
    • fix(tarko): scroll-to-bottom indicator session switching and edge cases (#1405) by @ulivz in 442dab8
    • fix(tarko): improve markdown link parsing edge cases (#1398) by @ulivz in 24fdf31

    Other Changes

    • refactor(tarko): remove excessive dots from empty states (#1414) by @ulivz in 074559e
    • chore(tars-stack): release 0.3.0-beta.10 by @ulivz in 59b59ef
    • chore: fix changelog filter scopes and restore missing entries (#1418) by @ulivz in b3fe00d
    • chore(o-agent): update example prompts (#1417) by @ulivz in 3b28c9b
    • chore(tarko): enhance code block spacing (#1400) by @ulivz in 1752459
    • chore(tars-stack): release 0.3.0-beta.9 (#1396) by @ulivz in da0e22b

    Full Changelog: v0.3.0-beta.9...v0.3.0-beta.10

用户点评

登录用户的真实使用记录,可按版本和硬件筛选
共 0 条

    还没有人点评。第一个说出真实体验的人最有价值。

    README 原文

    未翻译

    Introduction

    English | 简体中文

    TARS* is a Multimodal AI Agent stack, currently shipping two projects: Agent TARS and UI-TARS-desktop:

    Table of Contents

    News

    • [2025-11-05] 🎉 We're excited to announce the release of Agent TARS CLI v0.3.0! This version brings streaming support for multiple tools (shell commands, multi-file structured display), runtime settings with timing statistics for tool calls and deep thinking, Event Stream Viewer for data flow tracking and debugging. Additionally, it features exclusive support for AIO agent Sandbox as isolated all-in-one tools execution environment.
    • [2025-06-25] We released an Agent TARS Beta and Agent TARS CLI - Introducing Agent TARS Beta, a multimodal AI agent that aims to explore a work form that is closer to human-like task completion through rich multimodal capabilities (such as GUI Agent, Vision) and seamless integration with various real-world tools.
    • [2025-06-12] - 🎁 We are thrilled to announce the release of UI-TARS Desktop v0.2.0! This update introduces two powerful new features: Remote Computer Operator and Remote Browser Operator—both completely free. No configuration required: simply click to remotely control any computer or browser, and experience a new level of convenience and intelligence.
    • [2025-04-17] - 🎉 We're thrilled to announce the release of new UI-TARS Desktop application v0.1.0, featuring a redesigned Agent UI. The application enhances the computer using experience, introduces new browser operation features, and supports the advanced UI-TARS-1.5 model for improved performance and precise control.
    • [2025-02-20] - 📦 Introduced UI TARS SDK, is a powerful cross-platform toolkit for building GUI automation agents.
    • [2025-01-23] - 🚀 We updated the Cloud Deployment section in the 中文版: GUI模型部署教程 with new information related to the ModelScope platform. You can now use the ModelScope platform for deployment.

    Agent TARS

    Agent TARS is a general multimodal AI Agent stack, it brings the power of GUI Agent and Vision into your terminal, computer, browser and product. It primarily ships with a CLI and Web UI for usage. It aims to provide a workflow that is closer to human-like task completion through cutting-edge multimodal LLMs and seamless integration with various real-world MCP tools.

    Showcase

    Please help me book the earliest flight from San Jose to New York on September 1st and the last return flight on September 6th on Priceline
    

    https://github.com/user-attachments/assets/772b0eef-aef7-4ab9-8cb0-9611820539d8

    For more use cases, please check out #842.

    Core Features

    • 🖱️ One-Click Out-of-the-box CLI - Supports both headful Web UI and headless server execution.
    • 🌐 Hybrid Browser Agent - Control browsers using GUI Agent, DOM, or a hybrid strategy.
    • 🔄 Event Stream - Protocol-driven Event Stream drives Context Engineering and Agent UI.
    • 🧰 MCP Integration - The kernel is built on MCP and also supports mounting MCP Servers to connect to real-world tools.

    Quick Start

    # Launch with `npx`.
    npx @agent-tars/cli@latest
    
    # Install globally, required Node.js >= 22
    npm install @agent-tars/cli@latest -g
    
    # Run with your preferred model provider
    agent-tars --provider volcengine --model doubao-1-5-thinking-vision-pro-250428 --apiKey your-api-key
    agent-tars --provider anthropic --model claude-3-7-sonnet-latest --apiKey your-api-key
    

    Visit the comprehensive Quick Start guide for detailed setup instructions.

    Documentation

    🌟 Explore Agent TARS Universe 🌟

    UI-TARS Desktop

    UI-TARS Desktop is a native GUI agent for your local computer, driven by UI-TARS and Seed-1.5-VL/1.6 series models.

    Showcase

    InstructionLocal OperatorRemote Operator
    Please help me open the autosave feature of VS Code and delay AutoSave operations for 500 milliseconds in the VS Code setting.
    Could you help me check the latest open issue of the UI-TARS-Desktop project on GitHub?

    Features

    • 🤖 Natural language control powered by Vision-Language Model
    • 🖥️ Screenshot and visual recognition support
    • 🎯 Precise mouse and keyboard control
    • 💻 Cross-platform support (Windows/MacOS/Browser)
    • 🔄 Real-time feedback and status display
    • 🔐 Private and secure - fully local processing

    Quick Start

    See Quick Start

    Contributing

    See CONTRIBUTING.md.

    License

    This project is licensed under the Apache License 2.0.

    Citation

    If you find our paper and code useful in your research, please consider giving a star :star: and citation :pencil:

    @article{qin2025ui,
      title={UI-TARS: Pioneering Automated GUI Interaction with Native Agents},
      author={Qin, Yujia and Ye, Yining and Fang, Junjie and Wang, Haoming and Liang, Shihao and Tian, Shizuo and Zhang, Junda and Li, Jiahao and Li, Yunxin and Huang, Shijue and others},
      journal={arXiv preprint arXiv:2501.12326},
      year={2025}
    }