Kimi K3 est le mod\u00e8le de pointe ouvert de Moonshot AI, dot\u00e9 de 2,8 billions de param\u00e8tres \u2014 une architecture m\u00e9lange d\u2019experts
\nqui active uniquement 16 des 896 experts par jeton, avec une fen\u00eatre de contexte de 1 million de jetons, une prise en charge native
\nde la vision et un raisonnement permanent. Il obtient un score de 57 sur l\u2019Artificial Analysis Intelligence Index,
\nd\u00e9passant l\u00e9g\u00e8rement Claude Opus 4.8 (56) et ne c\u00e9dant que devant GPT-5.6 Sol (59) et Claude Fable 5 (60)
\n\u2014 le meilleur r\u00e9sultat open-weight jamais enregistr\u00e9 \u00e0 ce jour. Les poids ont \u00e9t\u00e9 publi\u00e9s le 27 juillet 2026.<\/p>\n
L\u2019importance r\u00e9side dans l\u2019\u00e9cart, ou plut\u00f4t dans son absence. Le fait qu\u2019un mod\u00e8le open-weight se situe
\n\u00e0 seulement trois points du meilleur mod\u00e8le ferm\u00e9 du march\u00e9 \u00e9limine le dernier argument commode
\njustifiant des prix \u00e9lev\u00e9s au motif exclusif des performances. \u00c0 3 $ en entr\u00e9e \/ 15 $ en sortie par
\nmillion de jetons, K3 co\u00fbte \u00e9galement deux fois moins cher que GPT-5.5 en co\u00fbt combin\u00e9, tout en obtenant un score sup\u00e9rieur. La difficult\u00e9 r\u00e9side
\ndans l\u2019exploitation : environ 1,4 To de VRAM en quantification 4 bits n\u00e9cessitent un cluster multi-n\u0153uds \u2014 deux n\u0153uds H200 \u00e0 8 GPU de 141 Go
\nou \u00e9quivalent \u2014 ainsi, \u00ab ouvert \u00bb signifie ici v\u00e9rifiable et portable entre fournisseurs, pas ex\u00e9cutable
\ndans votre salle serveur. Pour presque tous les utilisateurs, K3 constitue une d\u00e9cision d\u2019utilisation via API qui s\u2019accompagne toutefois
\nd\u2019une voie de sortie.<\/p>","protected":false},"excerpt":{"rendered":"
What is Kimi K3? Kimi K3 is Moonshot AI’s 2.8-trillion-parameter open frontier model \u2014 a mixture-of-experts that activates just 16 [\u2026]<\/p>\n","protected":false},"featured_media":2371,"template":"","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"cm_developer":"Moonshot AI","cm_model_type":"LLM (MoE, reasoning, multimodal)","cm_modality":"Text, Vision \u2192 Text","cm_parameters":"2.8T total \/ 16 of 896 experts active (MoE)","cm_context_window":"1M","cm_max_output":"","cm_license":"Open weights (due 2026-07-27)","cm_open_weights":"yes","cm_release_date":"2026-07","cm_input_price":"3.00","cm_output_price":"15.00","cm_api_providers":"Moonshot (Kimi), OpenRouter","cm_vram_fp16":"","cm_vram_q4":"~1.4 TB","cm_min_gpu":"Multi-node cluster (e.g. 2\u00d7 8-GPU H200 141GB nodes)","cm_benchmarks":"","cm_official_url":"https:\/\/www.kimi.com\/blog\/kimi-k3"},"class_list":["post-1570","ai_model","type-ai_model","status-publish","has-post-thumbnail","hentry"],"_links":{"self":[{"href":"https:\/\/convly.ai\/fr\/wp-json\/wp\/v2\/ai_model\/1570","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/convly.ai\/fr\/wp-json\/wp\/v2\/ai_model"}],"about":[{"href":"https:\/\/convly.ai\/fr\/wp-json\/wp\/v2\/types\/ai_model"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/convly.ai\/fr\/wp-json\/wp\/v2\/media\/2371"}],"wp:attachment":[{"href":"https:\/\/convly.ai\/fr\/wp-json\/wp\/v2\/media?parent=1570"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}