Priors Makes Time Charts Linear by Default

This is a sensible visualization fix: elapsed time is experienced linearly, so wall-clock differences should remain visually proportional. The change improves model-selection decisions without altering the underlying benchmark data.

• –Makes meaningful runtime gaps easier to spot when comparing models
• –Better reflects developer concerns around waiting for coding or agent tasks to finish
• –Keeps logarithmic scaling appropriate for cost comparisons spanning multiple orders of magnitude
• –Time-per-task still excludes first-token latency and orchestration overhead, so it is not equivalent to end-to-end user-perceived latency
• –Fits Priors’ broader role as a custom AI model-ranking and benchmark exploration tool