{"source":"Quelle IA, le classement des modèles de langage","page":"https://quelleia.com/benchmarks/deepresearch_bench/","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":"deepresearch_bench","nom":"DeepResearch Bench","tache":"agents","etat":"actif","echelle":"unite","modeles":24,"poids_general":1.67,"difficulte":92.6,"pente":0.0113,"en_une_phrase":"91 tâches de recherche sur le web : trouver un chiffre, compiler un jeu de données, vérifier une affirmation, retrouver une source.","notation":"Note moyenne de 0 à 1, selon la tâche (précision, rappel ou réponse juste), sur une copie figée du web et 50 actions au plus.","limites":"Les écarts entre modèles sont faibles, de 0,35 à 0,55. Le web figé ne reflète pas une recherche en conditions réelles.","mainteneur":"FutureSearch","url":"https://evals.futuresearch.ai/","verifie":true},"tache":{"id":"agents","nom":"Agents","poids_general":15,"benchmarks_actifs":9},"modeles_mesures":24,"mesures":41,"classement":[{"rang":1,"modele":"claude-opus-4-6","nom":"Claude Opus 4.6","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-6/","score_publie":0.5529999999999999,"score_0_1":0.553,"niveau_suggere":111.5,"reflexion":"elevee","harnais":null,"configurations_mesurees":3,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":2,"modele":"claude-sonnet-4-6","nom":"Claude Sonnet 4.6","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-sonnet-4-6/","score_publie":0.549,"score_0_1":0.549,"niveau_suggere":110.1,"reflexion":"elevee","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":3,"modele":"claude-opus-4-5","nom":"Claude Opus 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-5/","score_publie":0.5479999999999999,"score_0_1":0.548,"niveau_suggere":109.7,"reflexion":"elevee","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":4,"modele":"gpt-5-5","nom":"GPT-5.5","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-5/","score_publie":0.54,"score_0_1":0.54,"niveau_suggere":106.8,"reflexion":"elevee","harnais":null,"configurations_mesurees":3,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":5,"modele":"claude-sonnet-4-5","nom":"Claude Sonnet 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-sonnet-4-5/","score_publie":0.526,"score_0_1":0.526,"niveau_suggere":101.8,"reflexion":"budget","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":6,"modele":"claude-opus-4-8","nom":"Claude Opus 4.8","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-8/","score_publie":0.502,"score_0_1":0.502,"niveau_suggere":93.3,"reflexion":"elevee","harnais":null,"configurations_mesurees":3,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":7,"modele":"gemini-3-flash","nom":"Gemini 3 Flash","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-flash/","score_publie":0.498,"score_0_1":0.498,"niveau_suggere":91.9,"reflexion":"faible","harnais":null,"configurations_mesurees":3,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":8,"modele":"gpt-5","nom":"GPT-5","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5/","score_publie":0.496,"score_0_1":0.496,"niveau_suggere":91.2,"reflexion":"faible","harnais":null,"configurations_mesurees":4,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":9,"modele":"claude-opus-4-1","nom":"Claude Opus 4.1","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-1/","score_publie":0.483,"score_0_1":0.483,"niveau_suggere":86.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":10,"modele":"gemini-3-1-pro","nom":"Gemini 3.1 Pro","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-1-pro/","score_publie":0.478,"score_0_1":0.478,"niveau_suggere":84.8,"reflexion":"elevee","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":11,"modele":"grok-4","nom":"Grok 4","editeur":"xAI","page":"https://quelleia.com/modeles/grok-4/","score_publie":0.473,"score_0_1":0.473,"niveau_suggere":83,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":12,"modele":"claude-opus-4","nom":"Claude Opus 4","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4/","score_publie":0.4679999999999999,"score_0_1":0.468,"niveau_suggere":81.2,"reflexion":"budget","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":13,"modele":"claude-sonnet-4","nom":"Claude Sonnet 4","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-sonnet-4/","score_publie":0.466,"score_0_1":0.466,"niveau_suggere":80.5,"reflexion":"budget","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":14,"modele":"gemini-3-pro","nom":"Gemini 3 Pro","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-pro/","score_publie":0.4629999999999999,"score_0_1":0.463,"niveau_suggere":79.4,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":15,"modele":"claude-haiku-4-5","nom":"Claude Haiku 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-haiku-4-5/","score_publie":0.455,"score_0_1":0.455,"niveau_suggere":76.6,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":16,"modele":"o3","nom":"o3","editeur":"OpenAI","page":"https://quelleia.com/modeles/o3/","score_publie":0.452,"score_0_1":0.452,"niveau_suggere":75.5,"reflexion":"moyenne","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":17,"modele":"claude-3-7-sonnet","nom":"Claude 3.7 Sonnet","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-3-7-sonnet/","score_publie":0.436,"score_0_1":0.436,"niveau_suggere":69.8,"reflexion":"budget","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":18,"modele":"gpt-5-1","nom":"GPT-5.1","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-1/","score_publie":0.428,"score_0_1":0.428,"niveau_suggere":66.9,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":18,"modele":"gemini-2-5-pro-jun-2025","nom":"Gemini 2.5 Pro (Jun 2025)","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-2-5-pro-jun-2025/","score_publie":0.428,"score_0_1":0.428,"niveau_suggere":66.9,"reflexion":"inconnue","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":20,"modele":"gpt-5-2","nom":"GPT-5.2","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-2/","score_publie":0.411,"score_0_1":0.411,"niveau_suggere":60.7,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":21,"modele":"gemini-3-1-flash-lite","nom":"Gemini 3.1 Flash-Lite","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-1-flash-lite/","score_publie":0.373,"score_0_1":0.373,"niveau_suggere":46.5,"reflexion":"faible","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":22,"modele":"gpt-5-4-mini","nom":"GPT-5.4 Mini","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-4-mini/","score_publie":0.363,"score_0_1":0.363,"niveau_suggere":42.7,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":23,"modele":"gpt-5-4","nom":"GPT-5.4","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-4/","score_publie":0.351,"score_0_1":0.351,"niveau_suggere":38.1,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://evals.futuresearch.ai/#drb"},{"rang":23,"modele":"deepseek-r1-may-2025","nom":"DeepSeek-R1 (May 2025)","editeur":"DeepSeek","page":"https://quelleia.com/modeles/deepseek-r1-may-2025/","score_publie":0.351,"score_0_1":0.351,"niveau_suggere":38.1,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://epoch.ai/benchmarks"}]}