{"rewrite":{"id":"r_8a34cf957789c9a2903429d0","clusterId":"c_bb251ee76a4a7c8bb9dc2ad3","slug":"microsoft-releases-mai-code-2-and-mai-code-2-flash-for-github-copilot","model":"deepseek-v4-1-flash","headline":"Microsoft Releases MAI Code 2 and MAI Code 2 Flash for GitHub Copilot","summary":"Microsoft announced the MAI Code 2 and MAI Code 2 Flash programming models on November 17, 2026, and both are available in GitHub Copilot. Microsoft reports MAI Code 2 at a 63.2% average benchmark success rate against 64.1% for OpenAI's GPT-6.1 Sol and 65.3% for Anthropic's Claude Opus 5.5, at about 40% of GPT-6.1 Sol's per-task cost.","whyItMatters":"Microsoft is pricing its own coding model at roughly 40% of GPT-6.1 Sol's per-task cost while landing within a point of it on Microsoft's own benchmarks, which puts cost rather than raw accuracy at the center of the coding-model race.","webCardHtml":"\u003cp\u003eMicrosoft ran four benchmark tests to measure MAI Code 2\u0026#39;s programming ability. In a separate human evaluation, 258 participants compared screen designs and features, and rated MAI Code 2\u0026#39;s output superior to GPT-6 Sol\u0026#39;s in 64% of cases, judging both visual appearance and functionality.\u003c/p\u003e\u003cp\u003eMAI Code 2 Flash targets quick code explanations, partial fixes, and common bug fixes. Across three Microsoft benchmarks it averaged a 61.0% success rate, against 55.1% for GPT-6 Luna and 53.9% for Claude Haiku 5.5, at 65% of GPT-6 Luna\u0026#39;s processing cost and about 20% of Claude Haiku 5.5\u0026#39;s.\u003c/p\u003e\u003cp\u003eThe published cost figures use promotional pricing with a 50% discount, which runs until May 17, 2027. Microsoft also built MAI Work Flash on MAI Code 2 Flash, tuned for Excel, Word, PowerPoint, and Teams.\u003c/p\u003e","blueskyPost":"The benchmark spread between MAI Code 2, GPT-6.1 Sol, and Claude Opus 5.5 is under three points. Microsoft is selling the 40% cost figure, not the accuracy crown.","twitterPost":"MAI Code 2 trails GPT-6.1 Sol by 0.9 points. Microsoft's pitch is the 40% per-task cost, not the benchmark lead.","threadsPost":"MAI Code 2 lands at 63.2% against 64.1% for GPT-6.1 Sol and 65.3% for Claude Opus 5.5. Microsoft priced the story around 40% of GPT-6.1 Sol's per-task cost, so the near-tie matters less than the invoice.","newsletterBlurb":"Microsoft announced the MAI Code 2 and MAI Code 2 Flash programming models on November 17, 2026, and both are available in GitHub Copilot. Microsoft reports MAI Code 2 at a 63.2% average benchmark success rate, against 64.1% for GPT-6.1 Sol and 65.3% for Claude Opus 5.5, at about 40% of GPT-6.1 Sol's per-task cost. The cost figures use promotional pricing with a 50% discount that runs until May 17, 2027.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20261118-mai-code-2-flash/\",\"title\":\"Microsoft announces new AI programming models \\\"MAI Code 2\\\" and high-speed \\\"MAI Code 2 Flash,\\\" delivering performance comparable to GPT-6.1 Sol at 40% of the cost\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":5688,"outputTokens":989,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1794994831,"createdAt":"2026-11-18T09:36:18.000Z","publishedAt":"2026-11-18T09:39:16.000Z","updatedAt":"2026-11-18T09:39:16.000Z"},"cluster":{"id":"c_bb251ee76a4a7c8bb9dc2ad3","canonicalTitle":"Microsoftが新型AIプログラミングモデル「MAI Code 2」と高速版「MAI Code 2 Flash」を発表、GPT-6.1 Solに匹敵する性能を40％のコストで発揮","representativeArticleId":"a_02201b7063b9be81e94db130","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"MAI Code 2\",\"MAI Code 2 Flash\",\"GPT-6.1 Sol\",\"GitHub Copilot\"],\"studios\":[\"Microsoft\"],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-11-18T08:58:00.000Z","lastSeenAt":"2026-11-18T08:58:00.000Z","updatedAt":"2026-11-18T09:39:13.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20261118-mai-code-2-flash/","title":"Microsoftが新型AIプログラミングモデル「MAI Code 2」と高速版「MAI Code 2 Flash」を発表、GPT-6.1 Solに匹敵する性能を40％のコストで発揮"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["MAI Code 2","MAI Code 2 Flash","GPT-6.1 Sol","GitHub Copilot"],"studios":["Microsoft"],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":["Microsoft announced MAI Code 2 and MAI Code 2 Flash on November 17, 2026.","Both models are available in the GitHub Copilot programming assistance tool.","MAI Code 2 averaged a 63.2% success rate across four Microsoft benchmark tests, against 64.1% for GPT-6.1 Sol and 65.3% for Claude Opus 5.5.","MAI Code 2 Flash averaged 61.0% across three Microsoft benchmarks, against 55.1% for GPT-6 Luna and 53.9% for Claude Haiku 5.5.","The benchmark cost figures use promotional pricing with a 50% discount that runs until May 17, 2027."],"pageLinks":{"healthSlugs":["mai-code-2","mai-code-2-flash","gpt-6-1-sol","github-copilot","microsoft"],"storySlugs":[],"entities":[{"slug":"mai-code-2","supported":0,"disputed":0,"contradicted":0,"unverified":5,"overall":"unverified","story_count":1,"has_dossier":false},{"slug":"mai-code-2-flash","supported":0,"disputed":0,"contradicted":0,"unverified":3,"overall":"unverified","story_count":1,"has_dossier":false},{"slug":"gpt-6-1-sol","supported":0,"disputed":0,"contradicted":1,"unverified":3,"overall":"red","story_count":3,"has_dossier":true},{"slug":"github-copilot","supported":0,"disputed":0,"contradicted":0,"unverified":0,"overall":"unverified","story_count":1,"has_dossier":false},{"slug":"microsoft","supported":4,"disputed":0,"contradicted":0,"unverified":35,"overall":"green","story_count":36,"has_dossier":true}],"stories":[],"entitiesOk":true,"storiesOk":true}}
