Brand Hub
All templates
BENCH

8 Tasks Flipped on Harness Choice Alone · DeepSeek V4 Pro (0813)

Task × harness pass/fail matrix. Same model across all five harnesses, same eight tasks — but the pass column moves.

Download PNGOpen capture surface →
8 Tasks Flipped on Harness Choice Alone · DeepSeek V4 Pro (0813)

Template source: src/app/chart-templates/bench-harness-matrix/

Edit config.ts to change tasks, harnesses, or per-cell results. Fork this folder to make a new matrix template — add the slug to registry.ts.