Morgan(@morganlinton):Everyone asked me to do my next harness benchmark on Grok Build, heard you all loud and clear, working on getting everything setup now. Lots of little things to get right before testing a harness in VulcanBench. Harder than my normal benchmark setup, really have to make sure all strong guardrails are in place. I don't trust the honor system with models, need to know they can't cheat, not trust that they won't.

2026.08.17 16:47

Everyone asked me to do my next harness benchmark on Grok Build, heard you all loud and clear, working on getting everything setup now. Lots of little things to get right before testing a harness in VulcanBench. Harder than my normal benchmark setup, really have to make sure all strong guardrails are in place. I don't trust the honor system with models, need to know they can't cheat, not trust that they won't.

显示更多