Runta and Baseten find one model costs up to 5x more per task depending on the coding harness
Runta's FrontierHarness Eval v1.1 ran 22 coding agent harnesses with five open models across 27,280 runs. The tests use open models served by…
Topic
Runta's FrontierHarness Eval v1.1 ran 22 coding agent harnesses with five open models across 27,280 runs. The tests use open models served by…