{"source":"Quelle IA, le classement des modèles de langage","page":"https://quelleia.com/benchmarks/weirdml_v3/","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":"weirdml_v3","nom":"WeirdML v3","tache":"code","etat":"actif","echelle":"unite","modeles":11,"poids_general":1.15,"difficulte":102.6,"pente":0.1953,"en_une_phrase":"11 problèmes d'apprentissage automatique et d'analyse de données inhabituels : l'agent explore des données inconnues et construit un programme qui les traite.","notation":"Score de 0 à 1 qui récompense à la fois le résultat final et la rapidité à l'atteindre, mesurée en jetons consommés.","limites":"Onze modèles seulement à ce stade. Les scores ne se comparent pas à ceux de WeirdML v2, et le harnais (Claude Code, Codex CLI) varie selon le modèle.","mainteneur":"Håvard Tveit Ihle","url":"https://htihle.github.io/weirdml.html","verifie":true},"tache":{"id":"code","nom":"Code","poids_general":15,"benchmarks_actifs":13},"modeles_mesures":11,"mesures":11,"classement":[{"rang":1,"modele":"gpt-6-astra","nom":"GPT-6 Astra","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-6-astra/","score_publie":0.4222023136363637,"score_0_1":0.4222,"niveau_suggere":101,"reflexion":"maximale","harnais":"Codex CLI","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":2,"modele":"claude-opus-5-5","nom":"Claude Opus 5.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-5-5/","score_publie":0.3120028142857143,"score_0_1":0.312,"niveau_suggere":98.6,"reflexion":"maximale","harnais":"Claude Code","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":3,"modele":"claude-fable-5-1","nom":"Claude Fable 5.1","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-fable-5-1/","score_publie":0.2597021272727273,"score_0_1":0.2597,"niveau_suggere":97.3,"reflexion":"maximale","harnais":"Claude Code","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":4,"modele":"claude-opus-5","nom":"Claude Opus 5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-5/","score_publie":0.1901117537878787,"score_0_1":0.1901,"niveau_suggere":95.2,"reflexion":"maximale","harnais":"Claude Code","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":5,"modele":"claude-fable-5","nom":"Claude Fable 5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-fable-5/","score_publie":0.1552890606060606,"score_0_1":0.1553,"niveau_suggere":94,"reflexion":"maximale","harnais":"Claude Code","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":6,"modele":"gpt-5-6-sol","nom":"GPT-5.6 Sol","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-6-sol/","score_publie":0.1499222454545454,"score_0_1":0.1499,"niveau_suggere":93.7,"reflexion":"maximale","harnais":"Codex CLI","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":7,"modele":"gemini-3-8-flash","nom":"Gemini 3.8 Flash","editeur":"Google DeepMind","page":"https://quelleia.com/modeles/gemini-3-8-flash/","score_publie":0.0737874909090909,"score_0_1":0.0738,"niveau_suggere":89.7,"reflexion":"elevee","harnais":"Gemini CLI","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":8,"modele":"kimi-k3","nom":"Kimi K3","editeur":"Moonshot","page":"https://quelleia.com/modeles/kimi-k3/","score_publie":0.0726886515151515,"score_0_1":0.0727,"niveau_suggere":89.6,"reflexion":"elevee","harnais":"kimi_code","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":9,"modele":"claude-opus-4-5","nom":"Claude Opus 4.5","editeur":"Anthropic","page":"https://quelleia.com/modeles/claude-opus-4-5/","score_publie":0.06517475,"score_0_1":0.0652,"niveau_suggere":89,"reflexion":"elevee","harnais":"Claude Code","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":10,"modele":"deepseek-v4-1-flash","nom":"DeepSeek V4.1 Flash","editeur":"DeepSeek","page":"https://quelleia.com/modeles/deepseek-v4-1-flash/","score_publie":0.0594633181818181,"score_0_1":0.0595,"niveau_suggere":88.5,"reflexion":"elevee","harnais":"opencode","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"},{"rang":11,"modele":"gpt-5-6-luna","nom":"GPT-5.6 Luna","editeur":"OpenAI","page":"https://quelleia.com/modeles/gpt-5-6-luna/","score_publie":0.0515795227272727,"score_0_1":0.0516,"niveau_suggere":87.7,"reflexion":"maximale","harnais":"Codex CLI","configurations_mesurees":1,"date":null,"source":"https://htihle.github.io/weirdml.html"}]}