Compares total cost for this test against the test score. Quadrant lines are drawn at the median values. Only models with available cost data are shown.
12 low-scoring outliers hidden: Grok 4.20 (88.6%), Arcee AI: Trinity Mini (88.6%), Cydonia 24B V4.1 (87.3%), Ministral 3 3B (85.5%), Ministral 8B (80.9%), WizardLM 2 8x22b (79.1%), DeepSeek V4 Flash (77.7%), Ministral 3B (77.7%), Ministral 3 8B (77.3%), Thinking Machines Inkling (75.0%), Grok 4.20 (Reasoning) (72.3%), Cohere Command R+ (Aug. 2024) (68.2%).