Inception
ModelsModels from this lab.4
RankedModels with an index.1
Top modelModel with the highest index.47.5Mercury 2.5
LatestRelease of the newest model.8 Sept 2026Mercury 2.5
Best by category
Out of 100AgenticMulti-step tasks with tools.Mercury 2.550.4
CodingCode writing and repair.Mercury 2.543.8
ReasoningLogic problems and puzzles.Mercury 2.554.5
MultimodalTasks with images and text.—N/A
KnowledgeFacts and expert knowledge.Mercury 2.544.8
MultilingualTasks in many languages.—N/A
InstructionTasks with strict rules in the prompt.Mercury 2.554.4
MathMath problems.Mercury 2.549.7