RankingPresets

Presets

Choose a profile to rank models for your task. Weights show the relative importance of each capability.

PresetAgenticHard reasoningCodingHuman prefKnowledgeMultimodalLong context
Core work →Shared core profile: agentic work, reasoning and coding. Every selected capability must be supported.3433330000
Coding →Compare the coding capability only.001000000
Reasoning →Compare the reasoning capability only.010000000
Agentic →Compare the agentic capability only.100000000
Knowledge →Compare the knowledge capability only.000010000
Multimodal →Compare the multimodal capability only.000001000
Long context →Compare the long context capability only.000000100
Human pref →Compare the human pref capability only.000100000
All capabilities →Requires evidence for all seven capabilities. Partial profiles retain their capability estimates.14141614141414