{"rewrite":{"id":"r_79f46ba2d0292fd176075ede","clusterId":"c_1f5cd567d337a7c276bf7934","slug":"kimi-k3-surpasses-claude-fable-in-long-duration-coding-benchmarks","model":"deepseek-v4-flash","headline":"Kimi K3 Surpasses Claude Fable in Long-Duration Coding Benchmarks","summary":"Moonshot AI announced the Kimi K3 model on July 16, with 2.8 trillion parameters and a 1 million token context. It outperformed Claude Fable 5 and GPT 5.6 Sol on the SWE Marathon long-duration coding benchmark, scoring 42.0. On Terminal Bench 2.1 it beat Claude models but fell short of GPT 5.6 Sol. The model is open-weight, with weights to be released by July 27.","whyItMatters":"Kimi K3 shows that Chinese open-weight models are now competitive with frontier US models on long-duration coding tasks, while still trailing in debugging.","webCardHtml":"\u003cp\u003eMoonshot AI\u0026#39;s Kimi K3 uses a Mixture of Experts architecture with 896 specialized experts, activating 16 per input. The company claims it is the world\u0026#39;s first open model approaching 3 trillion parameters. Underlying technologies include Kimi Delta Attention for faster decoding in long contexts and Attention Residuals for training efficiency. The model also features native image understanding, enabling vision-in-the-loop for tasks like website design and game development. API pricing starts at $0.30 per million tokens for cached input.\u003c/p\u003e","blueskyPost":"Moonshot AI's Kimi K3, a 2.8 trillion parameter open model, beats Claude Fable 5 on long-duration coding benchmarks but trails in debugging. Weights due by July 27. #AI #KimiK3","twitterPost":"Moonshot AI's Kimi K3 (2.8T params, open-weight) surpasses Claude Fable 5 on SWE Marathon long-duration coding (42.0 vs 35.0). Trails in debugging. Weights release by July 27. #AI","threadsPost":null,"newsletterBlurb":"Moonshot AI released Kimi K3, a 2.8 trillion parameter open model that outperforms Claude Fable 5 on long-duration coding benchmarks. It uses a Mixture of Experts architecture and includes vision capabilities. Weights will be released by July 27.","attributionJson":"[{\"source\":\"ASCII.jp\",\"url\":\"https://ascii.jp/elem/000/004/420/4420382/?rss\",\"title\":\"DeepSeek Shock Returns? China's 'Kimi K3' Surpasses Claude Fable in Some Areas\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4299,"outputTokens":1903,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1784251408,"createdAt":"2026-07-17T01:22:06.000Z","publishedAt":"2026-07-17T01:22:28.000Z","updatedAt":"2026-07-17T01:22:06.000Z"},"cluster":{"id":"c_1f5cd567d337a7c276bf7934","canonicalTitle":"DeepSeekショック再来か　中国「Kimi K3」、一部でClaude Fable超え","representativeArticleId":"a_909150e3fb01dcef37294522","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"news\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-07-17T00:25:00.000Z","lastSeenAt":"2026-07-17T00:25:00.000Z","updatedAt":"2026-07-17T01:22:28.000Z"},"attribution":[{"source":"ASCII.jp","url":"https://ascii.jp/elem/000/004/420/4420382/?rss","title":"DeepSeekショック再来か　中国「Kimi K3」、一部でClaude Fable超え"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"news","domain":"other","is_roundup":false},"keyFacts":null}