Morgan(@morganlinton):Stop everything, benchmark Grok 4.6. And yeah, quite a few steps to get these benchmarks setup. I typically use Fable 5 Low effort as my orchestrator for these. As you can see, lots of updates need to be made to make sure pricing and effort are benchmarked correctly when a new model comes out.

Morgan

@morganlinton

Cofounder @BoldMetrics: the AI body data engine. Mad Scientist @VulcanBench: benchmarking models across effort levels on real coding tasks. Not an expert.

가입 January 2009

756 팔로잉 중 43K 팬

Morgan@morganlinton

2026.08.12 16:46

Stop everything, benchmark Grok 4.6. And yeah, quite a few steps to get these benchmarks setup. I typically use Fable 5 Low effort as my orchestrator for these. As you can see, lots of updates need to be made to make sure pricing and effort are benchmarked correctly when a new model comes out.