Arena Hard
Résultats des évaluations de l’IA et performances des modèles
Type: percentage
Résultats enregistrés
49
Score moyen
63.25%
Plage des scores
26.70% - 95.60%
Modèle en tête
95.60% - Qwen 3 235B A22B
Scores au fil du temps
Scores individuels du benchmark tracés par date.
Résultats des modèles
| Organisation | Modèles | Date du résultat | Meilleur score | Informations | Déclaré par le fournisseur | Source |
|---|---|---|---|---|---|---|
| Qwen 3 235B A22B | - | 95,60 % | - | Oui | Source | |
| Qwen 3 32B | - | 93,80 % | - | Oui | Source | |
| Qwen 3 30B A3B | - | 91 % | - | Oui | Source | |
| Qwen 3 30B A3B Instruct 2507 | - | 91 % | inferred version-family alias from qwen3-30b-a3b | Oui | Source | |
| Qwen 3 30B A3B Thinking 2507 | - | 91 % | inferred version-family alias from qwen3-30b-a3b | Oui | Source | |
| Qwen 3 Coder 30B A3B Instruct | - | 91 % | inferred high-confidence family alias from qwen3-30b-a3b (score=0.5007; benches=8) | Oui | Source | |
| Qwen 3 Omni 30B A3B Captioner | - | 91 % | inferred family alias from qwen3-30b-a3b (score=0.4129; benches=8) | Oui | Source | |
| Qwen 3 Omni 30B A3B Instruct | - | 91 % | inferred high-confidence family alias from qwen3-30b-a3b (score=0.4819; benches=8) | Oui | Source | |
| Qwen 3 Omni 30B A3B Thinking | - | 91 % | inferred high-confidence family alias from qwen3-30b-a3b (score=0.4819; benches=8) | Oui | Source | |
| Llama 3.3 Nemotron Super 49B v1 | 18 mars 2025 | 88,30 % | - | Oui | Source | |
| Llama 3.3 Nemotron Super 49B V1.5 | - | 88,30 % | inferred version-family alias from llama-3.3-nemotron-super-49b-v1 | Oui | Source | |
| Qwen 72B | - | 81,20 % | inferred family alias from qwen-2.5-72b-instruct (score=0.3060; benches=14) | Oui | Source | |
| Phi 4 Reasoning Plus | 30 avr. 2025 | 79 % | - | Oui | Source | |
| DeepSeek V2.5 (2024-09-05) | 05 sept. 2024 | 76,20 % | inferred alias from deepseek-v2.5 | Oui | Source | |
| DeepSeek V2.5 (2024-12-10) | 10 déc. 2024 | 76,20 % | inferred alias from deepseek-v2.5 | Oui | Source | |
| Phi 1 | - | 75,40 % | inferred family alias from phi-4 (score=0.3100; benches=13) | Oui | Source | |
| Phi 2 | - | 75,40 % | inferred family alias from phi-4 (score=0.3100; benches=13) | Oui | Source | |
| Phi 4 | 12 déc. 2024 | 75,40 % | - | Oui | Source | |
| Phi 4 Reasoning | 30 avr. 2025 | 73,30 % | - | Oui | Source | |
| Granite 4.1 30B | 29 avr. 2026 | 71,02 % | - | Oui | Source | |
| Granite 4.1 8B | 29 avr. 2026 | 68,98 % | - | Oui | Source | |
| Mistral Small 1.0 | 26 févr. 2024 | 58,30 % | inferred family alias from mistral-small-latest (score=0.3650; benches=9) | Oui | Source | |
| Mistral Small 2.0 | 17 sept. 2024 | 58,30 % | inferred family alias from mistral-small-latest (score=0.3650; benches=9) | Oui | Source | |
| Mistral Small 4 | 16 mars 2026 | 58,30 % | - | Oui | Source | |
| Mistral Small Creative | 16 déc. 2025 | 58,30 % | inferred family alias from mistral-small-latest (score=0.4273; benches=9) | Oui | Source |
