Alternative to artificialanalysis for assessing latest models?
retained text
AA no longer lists GLM5.3 flash (despite clearly being one of the best models - and probably the cheapest per intelligence) , they seem to have been doing all sorts of changes to push US models and hiding good cost effective models, including qwen 3.8 flash, so that table is no good. I need a better one. Any alternatives.
document fetched 2026-10-29 00:44 · canonical https://reddit.com/r/LocalLLaMA/comments/1xhqlrw/alternative_to_artificialanalysis_for_assessing