@skeptrune Both are in decline tbh. It's an outdated category that only worked with the limited pre-opus 4.5 models. Codex, CC and Opencode are the only ones that matter now.
@s0me1suspicious@scaling01 Deepseek flash is extremely fast inference. As far as I know only Gemini flash and cerebras hosted models are faster at similar quality
@cjwynes@Drachs1978 You think the people who made the greatest invention of this century are going to rather do a career change to some meaningless job than move to Europe, Singapore, Australia, Japan, etc?