{"source":"Quelle IA, le classement des modèles de langage","page":"https://quelleia.com/benchmarks/osworld/","edition":"2026-W40","date":"2026-09-28","echelle":"score_publie est le score de la source, dans son unité. score_0_1 est ce score ramené entre 0 et 1 pour le calcul. niveau_suggere est le score IA que cette seule mesure indique : 100 pour le meilleur modèle au lancement du site, 50 pour GPT-4o de novembre 2024. difficulte est le score IA d'un modèle qui obtient la moitié (score_0_1 = 0,5), pente la raideur de la courbe.","licence":"Réutilisation libre en citant « Quelle IA, quelleia.com » et la date de l'édition. Les mesures renvoient à leurs sources d'origine.","benchmark":{"id":"osworld","nom":"OSWorld","tache":"agents","etat":"archive","echelle":"pourcent","modeles":9,"poids_general":0,"difficulte":75.4,"pente":0.099,"en_une_phrase":"Accomplir des tâches sur un vrai ordinateur : manipuler des fichiers, régler une application, enchaîner plusieurs logiciels.","notation":"Pourcentage des 369 tâches réussies, vérifié par script sur l'état final de la machine.","limites":"Le score dépend beaucoup du harnais et du nombre d'étapes autorisées. Les meilleurs agents rejoignent le niveau humain mesuré, environ 72 %.","mainteneur":"XLANG Lab (université de Hong Kong)","url":"https://os-world.github.io/","verifie":true},"tache":{"id":"agents","nom":"Agents","poids_general":15,"benchmarks_actifs":9},"modeles_mesures":9,"mesures":20,"classement":[{"rang":1,"modele":"claude-sonnet-4-6","nom":"Claude Sonnet 4.6","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-sonnet-4-6/","score_publie":72.1,"score_0_1":0.721,"niveau_suggere":85,"reflexion":"inconnue","harnais":"claude-sonnet-4-6 (100 steps)","configurations_mesurees":1,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":2,"modele":"claude-opus-4-5","nom":"Claude Opus 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-5/","score_publie":66.3,"score_0_1":0.663,"niveau_suggere":82.2,"reflexion":"inconnue","harnais":"Claude Opus 4.5","configurations_mesurees":1,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":3,"modele":"kimi-k2-5","nom":"Kimi K2.5","editeur":"Moonshot","page":"https://quelleia.com/modeles/kimi-k2-5/","score_publie":63.3,"score_0_1":0.633,"niveau_suggere":80.9,"reflexion":"inconnue","harnais":"Kimi-K2.5","configurations_mesurees":1,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":4,"modele":"claude-sonnet-4-5","nom":"Claude Sonnet 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-sonnet-4-5/","score_publie":62.9,"score_0_1":0.629,"niveau_suggere":80.7,"reflexion":"inconnue","harnais":"claude-sonnet-4-5-20250929 (100 steps)","configurations_mesurees":3,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":5,"modele":"claude-sonnet-4","nom":"Claude Sonnet 4","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-sonnet-4/","score_publie":43.9,"score_0_1":0.439,"niveau_suggere":72.9,"reflexion":"inconnue","harnais":"claude-4-sonnet-20250514 (50 steps)","configurations_mesurees":3,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":6,"modele":"claude-3-7-sonnet","nom":"Claude 3.7 Sonnet","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-3-7-sonnet/","score_publie":35.8,"score_0_1":0.358,"niveau_suggere":69.5,"reflexion":"inconnue","harnais":"claude-3-7-sonnet-20250219 (50 steps)","configurations_mesurees":3,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":7,"modele":"computer-use-preview-2025-03-11","nom":"computer-use-preview-2025-03-11","editeur":"OpenAI","page":"https://quelleia.com/modeles/computer-use-preview-2025-03-11/","score_publie":31.3,"score_0_1":0.313,"niveau_suggere":67.4,"reflexion":"inconnue","harnais":"computer-use-preview (50 steps)","configurations_mesurees":3,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":8,"modele":"o3","nom":"o3","editeur":"OpenAI","page":"https://quelleia.com/modeles/o3/","score_publie":23,"score_0_1":0.23,"niveau_suggere":63.2,"reflexion":"moyenne","harnais":"o3 (100 steps)","configurations_mesurees":3,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":9,"modele":"qwen2-5-72b","nom":"Qwen2.5-72B","editeur":"Alibaba","page":"https://quelleia.com/modeles/qwen2-5-72b/","score_publie":5,"score_0_1":0.05,"niveau_suggere":45.6,"reflexion":"inconnue","harnais":"qwen2.5-vl-72b-instruct (100 steps)","configurations_mesurees":2,"date":null,"source":"https://epoch.ai/benchmarks"}]}