{"source":"Quelle IA, le classement des modèles de langage","page":"https://quelleia.com/benchmarks/balrog/","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":"balrog","nom":"BALROG","tache":"agents","etat":"actif","echelle":"unite","modeles":37,"poids_general":1.67,"difficulte":85.1,"pente":0.0372,"en_une_phrase":"Jouer à six jeux, du simple BabyAI au redoutable NetHack, qui demandent d'explorer, de planifier et de tenir sur la durée.","notation":"Progression moyenne dans les six jeux, de 0 à 1, calculée sur plusieurs parties par jeu.","limites":"La moyenne cache de grands écarts : les modèles terminent les jeux simples mais avancent à peine dans NetHack.","mainteneur":"Équipe BALROG (auteurs de l'article de 2024)","url":"https://balrogai.com/","verifie":true},"tache":{"id":"agents","nom":"Agents","poids_general":15,"benchmarks_actifs":9},"modeles_mesures":37,"mesures":41,"classement":[{"rang":1,"modele":"gpt-6-astra","nom":"GPT-6 Astra","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-6-astra/","score_publie":0.6829999999999999,"score_0_1":0.683,"niveau_suggere":105.7,"reflexion":"maximale","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":2,"modele":"claude-opus-5","nom":"Claude Opus 5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-5/","score_publie":0.634,"score_0_1":0.634,"niveau_suggere":99.8,"reflexion":"maximale","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":3,"modele":"gpt-5-6-sol","nom":"GPT-5.6 Sol","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-6-sol/","score_publie":0.6,"score_0_1":0.6,"niveau_suggere":96,"reflexion":"maximale","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":4,"modele":"gemini-3-pro","nom":"Gemini 3 Pro","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-pro/","score_publie":0.5810000000000001,"score_0_1":0.581,"niveau_suggere":93.9,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":5,"modele":"gemini-3-1-pro","nom":"Gemini 3.1 Pro","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-1-pro/","score_publie":0.57,"score_0_1":0.57,"niveau_suggere":92.7,"reflexion":"inconnue","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":6,"modele":"gpt-5-6-terra","nom":"GPT-5.6 Terra","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-6-terra/","score_publie":0.532,"score_0_1":0.532,"niveau_suggere":88.5,"reflexion":"maximale","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":7,"modele":"gemini-3-flash","nom":"Gemini 3 Flash","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-flash/","score_publie":0.481,"score_0_1":0.481,"niveau_suggere":83,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":8,"modele":"gpt-5-6-luna","nom":"GPT-5.6 Luna","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-6-luna/","score_publie":0.456,"score_0_1":0.456,"niveau_suggere":80.3,"reflexion":"maximale","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":9,"modele":"grok-4","nom":"Grok 4","editeur":"xAI","page":"https://quelleia.com/modeles/grok-4/","score_publie":0.436,"score_0_1":0.436,"niveau_suggere":78.2,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":10,"modele":"claude-opus-4-5","nom":"Claude Opus 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-5/","score_publie":0.435,"score_0_1":0.435,"niveau_suggere":78,"reflexion":"inconnue","harnais":null,"configurations_mesurees":3,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":11,"modele":"gemini-2-5-pro-mar-2025","nom":"Gemini 2.5 Pro (Mar 2025)","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-2-5-pro-mar-2025/","score_publie":0.433,"score_0_1":0.433,"niveau_suggere":77.8,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":12,"modele":"deepseek-r1","nom":"DeepSeek-R1","editeur":"DeepSeek","page":"https://quelleia.com/modeles/deepseek-r1/","score_publie":0.349,"score_0_1":0.349,"niveau_suggere":68.3,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":13,"modele":"gemini-2-5-flash-jun-2025","nom":"Gemini 2.5 Flash (Jun 2025)","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-2-5-flash-jun-2025/","score_publie":0.335,"score_0_1":0.335,"niveau_suggere":66.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":14,"modele":"gpt-5","nom":"GPT-5","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5/","score_publie":0.3279999999999999,"score_0_1":0.328,"niveau_suggere":65.8,"reflexion":"faible","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":15,"modele":"claude-3-5-sonnet-october-2024","nom":"Claude 3.5 Sonnet (October 2024)","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-3-5-sonnet-october-2024/","score_publie":0.326,"score_0_1":0.326,"niveau_suggere":65.5,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":16,"modele":"gpt-4o-may-2024","nom":"GPT-4o (May 2024)","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-4o-may-2024/","score_publie":0.3229999999999999,"score_0_1":0.323,"niveau_suggere":65.2,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":17,"modele":"claude-haiku-4-5","nom":"Claude Haiku 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-haiku-4-5/","score_publie":0.312,"score_0_1":0.312,"niveau_suggere":63.8,"reflexion":"inconnue","harnais":null,"configurations_mesurees":2,"date":null,"source":"https://epoch.ai/benchmarks"},{"rang":18,"modele":"grok-3","nom":"Grok 3","editeur":"xAI","page":"https://quelleia.com/modeles/grok-3/","score_publie":0.295,"score_0_1":0.295,"niveau_suggere":61.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":19,"modele":"reka-flash-3","nom":"Reka Flash 3","editeur":"Reka AI","page":"https://quelleia.com/modeles/reka-flash-3/","score_publie":0.292,"score_0_1":0.292,"niveau_suggere":61.3,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":20,"modele":"llama-3-1-70b","nom":"Llama 3.1-70B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-1-70b/","score_publie":0.2789999999999999,"score_0_1":0.279,"niveau_suggere":59.5,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":21,"modele":"llama-3-2-90b","nom":"Llama 3.2 90B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-2-90b/","score_publie":0.273,"score_0_1":0.273,"niveau_suggere":58.7,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":22,"modele":"llama-3-3-70b","nom":"Llama 3.3 70B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-3-70b/","score_publie":0.23,"score_0_1":0.23,"niveau_suggere":52.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":23,"modele":"gemini-1-5-pro-sept-2024","nom":"Gemini 1.5 Pro (Sept 2024)","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-1-5-pro-sept-2024/","score_publie":0.21,"score_0_1":0.21,"niveau_suggere":49.5,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":24,"modele":"deepseek-r1-distill-qwen-32b","nom":"DeepSeek-R1-Distill-Qwen-32B","editeur":"DeepSeek","page":"https://quelleia.com/modeles/deepseek-r1-distill-qwen-32b/","score_publie":0.195,"score_0_1":0.195,"niveau_suggere":47,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":25,"modele":"claude-3-5-haiku","nom":"Claude 3.5 Haiku","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-3-5-haiku/","score_publie":0.193,"score_0_1":0.193,"niveau_suggere":46.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":26,"modele":"mistral-nemo","nom":"Mistral NeMo","editeur":"Mistral AI","page":"https://quelleia.com/modeles/mistral-nemo/","score_publie":0.176,"score_0_1":0.176,"niveau_suggere":43.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":27,"modele":"gpt-4o-mini","nom":"GPT-4o mini","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-4o-mini/","score_publie":0.174,"score_0_1":0.174,"niveau_suggere":43.2,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":28,"modele":"llama-3-2-11b","nom":"Llama 3.2 11B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-2-11b/","score_publie":0.168,"score_0_1":0.168,"niveau_suggere":42.1,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":29,"modele":"qwen2-5-72b","nom":"Qwen2.5-72B","editeur":"Alibaba","page":"https://quelleia.com/modeles/qwen2-5-72b/","score_publie":0.162,"score_0_1":0.162,"niveau_suggere":40.9,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":30,"modele":"llama-3-1-8b","nom":"Llama 3.1-8B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-1-8b/","score_publie":0.151,"score_0_1":0.151,"niveau_suggere":38.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":31,"modele":"gemini-1-5-flash-sep-2024","nom":"Gemini 1.5 Flash (Sep 2024)","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-1-5-flash-sep-2024/","score_publie":0.146,"score_0_1":0.146,"niveau_suggere":37.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":32,"modele":"qwen2-vl-72b-instruct","nom":"Qwen2-VL-72B-Instruct","editeur":"Alibaba","page":"https://quelleia.com/modeles/qwen2-vl-72b-instruct/","score_publie":0.128,"score_0_1":0.128,"niveau_suggere":33.5,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":33,"modele":"phi-4","nom":"Phi-4","editeur":"Microsoft","page":"https://quelleia.com/modeles/phi-4/","score_publie":0.1159999999999999,"score_0_1":0.116,"niveau_suggere":30.5,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":34,"modele":"llama-3-2-3b","nom":"Llama 3.2 3B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-2-3b/","score_publie":0.1009999999999999,"score_0_1":0.101,"niveau_suggere":26.3,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":35,"modele":"qwen2-5-7b","nom":"Qwen2.5-7B","editeur":"Alibaba","page":"https://quelleia.com/modeles/qwen2-5-7b/","score_publie":0.078,"score_0_1":0.078,"niveau_suggere":18.7,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":36,"modele":"llama-3-2-1b","nom":"Llama 3.2 1B","editeur":"Meta AI","page":"https://quelleia.com/modeles/llama-3-2-1b/","score_publie":0.066,"score_0_1":0.066,"niveau_suggere":13.8,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"},{"rang":37,"modele":"qwen2-vl-7b-instruct","nom":"Qwen2-VL-7B-Instruct","editeur":"Alibaba","page":"https://quelleia.com/modeles/qwen2-vl-7b-instruct/","score_publie":0.037,"score_0_1":0.037,"niveau_suggere":-2.6,"reflexion":"inconnue","harnais":null,"configurations_mesurees":1,"date":null,"source":"https://balrogai.com/"}]}