[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f6b0f4ve4bnk8":3,"$f2f16xg4r9og3w":23,"$fp95wgtyv7hyu":25,"$f3bozajl8yebe7":29,"$f8ypa021er9qz":42,"$f16awueqtcifs6":44},{"id":4,"name":5,"publisher":6,"paramsB":7,"useCase":8,"licenseStatus":9,"licenseUrl":7,"sourceUrl":10,"descriptionZh":11,"descriptionEn":12,"imageUrl":7,"downloadUrls":13,"stats":14,"variants":17,"derivations":18,"measuredHardware":21,"updatedAt":22},91,"MiniMax-M3.1-Flash","MiniMax",null,"CHAT","RESTRICTED","https:\u002F\u002Fwww.minimax.io\u002Fblog\u002Fminimax-m31f","MiniMax 于 2026-10-19 发布的 Flash 系列新旗舰（API 先行，权重后置）。相对前代更激进的稀疏化：官方口径在线推理约 3 倍速、任务成本降 70%（$0.15 输入 \u002F $0.60 输出 \u002F $0.015 缓存输入，每百万 token）；输出 200+ tok\u002Fs（P50 170+）。基准：BankerToolBench 76.2、DeepSWE v1.1 73.7%、OSWorld V2 61.7、Agents Last Exam 29.6（中国模型双第一）、CyberGym 82%。匿名声测期代号 Space Bunny，已处理 200 万亿 token。当前仅 MiniMax API \u002F M Plan \u002F MiniMax Code 可用，官方称模型权重稍后放出——参数量未公布，本地实测待权重发布。","MiniMax's new Flash-series flagship (Oct 19, API-first with weights to follow). More aggressive sparsity than the predecessor: official claims ~3x faster online inference and 70% lower task cost ($0.15 input \u002F $0.60 output \u002F $0.015 cached per 1M tokens), with 200+ tok\u002Fs output (P50 170+). Benchmarks: BankerToolBench 76.2, DeepSWE v1.1 73.7%, OSWorld V2 61.7, Agents Last Exam 29.6 (first among Chinese models on both), CyberGym 82%. Ran anonymously as \"Space Bunny\" processing 200 trillion tokens. Available now via MiniMax API, M Plan and MiniMax Code; model weights promised later — parameter count unpublished, local runs await the weights.",[],{"recordCount":15,"variantCount":15,"hardwareCount":15,"unlinkedRecordCount":15,"evidence":16},0,{"L0":15,"L1":15,"L2":15},[],{"parents":19,"children":20},[],[],[],"2026-10-20T11:57:47.032Z",{"favorited":24},false,{"items":26,"total":15,"page":27,"pageSize":28},[],1,50,{"items":30,"total":27,"page":27,"pageSize":41},[31],{"id":32,"type":33,"title":34,"excerpt":35,"author":36,"replyCount":38,"createdAt":39,"lastRepliedAt":7,"resolvedAt":7,"editedAt":7,"model":40,"hardware":7,"variant":7,"record":7},36,"HELP","MiniMax-M3.1-Flash：API 先上还是等权重？算了下和 Flash 三兄弟的成本账","M3.1-Flash 昨天发布，API 已可用（$0.15 \u002F $0.60 \u002F $0.015 每百万 token），权重说「稍后放出」。站内已按 RESTRICTED 口径收录（参数量都没公布）。 我的场景是长对话 agent（日均 ~5M",{"nickname":37,"avatarUrl":7},"江南卡皇",3,"2026-10-20T12:21:10.405Z",{"name":5},20,{"items":43},[],{"items":45,"total":27,"page":27,"pageSize":38},[46],{"slug":47,"category":48,"publishedAt":49,"titleZh":50,"titleEn":51,"summaryZh":52,"summaryEn":53,"models":54,"hardwares":65},"minimax-m31-flash-2026-10-20","NEWS","2026-10-20T10:00:00.000Z","MiniMax-M3.1-Flash 发布：API 先行、权重后置，在线 3 倍速、成本降 70%","MiniMax-M3.1-Flash ships: API first, weights later, 3x faster online at 70% lower cost","10 月 19 日 MiniMax 上线 M3.1-Flash（Flash 系列新旗舰，API 先行）。官方口径：在线推理约 3 倍速、任务成本降 70%（$0.15 \u002F $0.60 \u002F $0.015 每百万 token），输出 200+ tok\u002Fs。匿名声测代号 Space Bunny，已处理 200 万亿 token。权重稍后放出——站内按 RESTRICTED \u002F 参数未公布口径收录，零实测记录。","MiniMax shipped M3.1-Flash on Oct 19, the new Flash-series flagship, API first. Official claims: roughly 3x faster online inference, 70% lower task cost ($0.15 \u002F $0.60 \u002F $0.015 per 1M tokens), 200+ tok\u002Fs output. It ran anonymously as Space Bunny, processing 200 trillion tokens. Weights come later — logged as RESTRICTED with unpublished params, zero on-site runs.",[55,58,61,64],{"id":56,"name":57},9,"GLM-5.3-Flash",{"id":59,"name":60},11,"Qwen3.8-Flash-Next",{"id":62,"name":63},22,"MiniMax-M3",{"id":4,"name":5},[]]