{"id":7333,"date":"2025-09-03T06:16:00","date_gmt":"2025-09-02T21:16:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7333"},"modified":"2025-09-01T13:58:06","modified_gmt":"2025-09-01T04:58:06","slug":"livemcp-101-mcp-universe","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7333","title":{"rendered":"LiveMCP-101, MCP-Universe, MCP-Bench"},"content":{"rendered":"\n<p>MCP\u306b\u95a2\u3059\u308b\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u304c\u3067\u3066\u3044\u305f\u3002\u4e21\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3068\u3082GPT-5\u306e\u6027\u80fd\u304c\u9ad8\u3044\u3068\u306e\u3053\u3068\u3060\u304c\u3001\u591a\u304f\u306eMCP\u30b5\u30fc\u30d0\u3084\u5468\u8fba\u30c4\u30fc\u30eb\u30fb\u30e9\u30a4\u30d6\u30e9\u30ea\u304cGPT-4\/4.1\/4.5\/5\u306a\u3069\u306b\u5bfe\u3057\u3066\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u3066\u3044\u308b\u9762\u3082\u3042\u308b\u3088\u3046\u306b\u601d\u308f\u306a\u304f\u3082\u306a\u3044\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>LiveMCP-101: Stress Testing and Diagnosing MCP-enabled Agents on Challenging Queries&nbsp;<\/strong>[38.6]<br>\u63d0\u6848\u3059\u308bLiveMCP-101\u306f,\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u30af\u30a8\u30ea\u3092\u614e\u91cd\u306b\u30ad\u30e5\u30ec\u30fc\u30c8\u3057\u305f101\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u3042\u308b\u3002 \u5b9f\u9a13\u306b\u3088\u308a\u3001\u30d5\u30ed\u30f3\u30c6\u30a3\u30a2\u306eLLM\u3067\u3055\u304860%\u672a\u6e80\u306e\u6210\u529f\u7387\u3092\u9054\u6210\u3059\u308b\u3053\u3068\u304c\u793a\u3055\u308c\u305f\u3002 LiveMCP-101\u306f\u73fe\u5b9f\u4e16\u754c\u306e\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u80fd\u529b\u3092\u8a55\u4fa1\u3059\u308b\u305f\u3081\u306e\u53b3\u683c\u306a\u6a19\u6e96\u3092\u8a2d\u5b9a\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2508.15760v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2508.15760v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Thu, 21 Aug 2025 17:55:54 GMT)<\/li>\n\n\n\n<li>\u300cwe present LiveMCP-101, a benchmark of 101 carefully curated real-world queries, refined through iterative LLM rewriting and manual review, that require coordinated use of multiple MCP tools including web search, file operations, mathematical reasoning, and data analysis.\u300d\u3068\u3044\u3046\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3002<\/li>\n\n\n\n<li>GPT-5\u306e\u6027\u80fd\u304c\u9ad8\u3044\u3002<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>MCP-Universe: Benchmarking Large Language Models with Real-World Model Context Protocol Servers&nbsp;<\/strong>[86.0]<br>MCP-Universe\u306f,\u5b9f\u4e16\u754c\u306eMPP\u30b5\u30fc\u30d0\u3068\u306e\u30a4\u30f3\u30bf\u30e9\u30af\u30b7\u30e7\u30f3\u3092\u901a\u3058\u3066,\u73fe\u5b9f\u7684\u304b\u3064\u56f0\u96e3\u306a\u30bf\u30b9\u30af\u306b\u304a\u3044\u3066LLM\u3092\u8a55\u4fa1\u3059\u308b\u305f\u3081\u306b\u8a2d\u8a08\u3055\u308c\u305f,\u521d\u3081\u3066\u306e\u7dcf\u5408\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u3042\u308b\u3002 \u79c1\u305f\u3061\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u306f\u3001\u30ed\u30b1\u30fc\u30b7\u30e7\u30f3\u30ca\u30d3\u30b2\u30fc\u30b7\u30e7\u30f3\u3001\u30ea\u30dd\u30b8\u30c8\u30ea\u7ba1\u7406\u3001\u8ca1\u52d9\u5206\u6790\u30013D\u30c7\u30b6\u30a4\u30f3\u3001\u30d6\u30e9\u30a6\u30b6\u81ea\u52d5\u5316\u3001Web\u691c\u7d22\u3068\u3044\u3046\u300111\u306e\u7570\u306a\u308bMSP\u30b5\u30fc\u30d0\u306b\u307e\u305f\u304c\u308b6\u3064\u306e\u30b3\u30a2\u30c9\u30e1\u30a4\u30f3\u3092\u7db2\u7f85\u3057\u3066\u3044\u307e\u3059\u3002 GPT-5 (43.72%) \u3084Grok-4 (33.33%) \u3084Claude-4.0-Sonnet (29.44%) \u306e\u3088\u3046\u306aSOTA\u30e2\u30c7\u30eb\u3067\u3055\u3048\u3001\u5927\u5e45\u306a\u6027\u80fd\u5236\u9650\u304c\u3042\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2508.14704v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2508.14704v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Wed, 20 Aug 2025 13:28:58 GMT)<\/li>\n\n\n\n<li>\u3053\u3061\u3089\u3082\u300cwe introduce MCP-Universe, the first comprehensive benchmark specifically designed to evaluate LLMs in realistic and hard tasks through interaction with real-world MCP servers. Our benchmark encompasses 6 core domains spanning 11 different MCP servers: Location Navigation, Repository Management, Financial Analysis, 3D Design, Browser Automation, and Web Searching.\u300d\u3068MCP\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3001GPT-5\u306e\u6027\u80fd\u304c\u9ad8\u3044\u3002<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f<a href=\"https:\/\/mcp-universe.github.io\/\">MCP-Universe: Benchmarking Large Language Models with Real-World Model Context Protocol Servers<\/a>\u3001\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/SalesforceAIResearch\/MCP-Universe\">GitHub &#8211; SalesforceAIResearch\/MCP-Universe: MCP-Universe is a comprehensive framework designed for developing, testing, and benchmarking AI agents<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers\u00a0<\/strong>[24.7]<br>MCP-Bench\u306f\u3001\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb(LLM)\u3092\u73fe\u5b9f\u7684\u306a\u30de\u30eb\u30c1\u30b9\u30c6\u30c3\u30d7\u30bf\u30b9\u30af\u3067\u8a55\u4fa1\u3059\u308b\u305f\u3081\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u3042\u308b\u3002 MCP-Bench \u306f Model Context Protocol (MCP) \u4e0a\u306b\u69cb\u7bc9\u3055\u308c\u3066\u304a\u308a\u3001\u91d1\u878d\u3001\u65c5\u884c\u3001\u79d1\u5b66\u8a08\u7b97\u3001\u5b66\u8853\u691c\u7d22\u306a\u3069\u306e\u5206\u91ce\u306b\u307e\u305f\u304c\u308b250\u306e\u30c4\u30fc\u30eb\u306b\u307e\u305f\u304c\u308b28\u306e\u30e9\u30a4\u30d6 MCP \u30b5\u30fc\u30d0\u306b LLM \u3092\u63a5\u7d9a\u3057\u3066\u3044\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2508.20453v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2508.20453v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 28 Aug 2025 05:58:57 GMT)<\/li>\n\n\n\n<li>\u30a2\u30af\u30bb\u30f3\u30c1\u30e5\u30a2\u306b\u3088\u308b\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3002GPT-5\u3001o3\u3001GPT-OSS 120B\u3001Gemini 2.5 Pro\u3001Claude sonnet 4\u3068\u7d9a\u304f\u7d50\u679c\u3002\n<ul class=\"wp-block-list\">\n<li>\u611f\u899a\u3068\u304b\u306a\u308a\u7570\u306a\u308b\u5370\u8c61\u3067MCP\u30b5\u30fc\u30d0\u5074\u304cGPT\u7cfb\u30e2\u30c7\u30eb\u306b\u5bc4\u305b\u3066\u3044\u308b\u6c17\u304c\u3059\u308b<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/Accenture\/mcp-bench\">GitHub &#8211; Accenture\/mcp-bench: MCP-Bench: Benchmarking Tool-Using LLM Agents with Complex Real-World Tasks via MCP Servers<\/a><\/li>\n<\/ul>\n\n\n\n<p><\/p>\n","protected":false},"excerpt":{"rendered":"<p>MCP\u306b\u95a2\u3059\u308b\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u304c\u3067\u3066\u3044\u305f\u3002\u4e21\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3068\u3082GPT-5\u306e\u6027\u80fd\u304c\u9ad8\u3044\u3068\u306e\u3053\u3068\u3060\u304c\u3001\u591a\u304f\u306eMCP\u30b5\u30fc\u30d0\u3084\u5468\u8fba\u30c4\u30fc\u30eb\u30fb\u30e9\u30a4\u30d6\u30e9\u30ea\u304cGPT-4\/4.1\/4.5\/5\u306a\u3069\u306b\u5bfe\u3057\u3066\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3055\u308c\u3066\u3044\u308b\u9762\u3082\u3042\u308b\u3088\u3046\u306b\u601d\u308f\u306a &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=7333\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;LiveMCP-101, MCP-Universe, MCP-Bench&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[695,517],"class_list":["post-7333","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-mcp","tag-517"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7333","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=7333"}],"version-history":[{"count":3,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7333\/revisions"}],"predecessor-version":[{"id":7371,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7333\/revisions\/7371"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7333"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7333"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7333"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}