> 80-90% efficiency
replywdym by that
> for daily tasks
which are?
You get about 80-90% of the results for daily tasks like: getting summaries or explanations of complex material. Writing software tools for data analysis. Getting recipes for a given set of ingredients in the fridge.
reply128B-A16B class models at 10-50 tok/s should be plenty for most tasks done on computers
replyWhat do you need a frontier model for, really.
replyFrontier models like Fable are mostly useful if you want to paste in one or two prompts, and receive a subtly broken application that looks impressive. That is very hard to do with local models today.
replyWhat current local models work fine for is delegating clearly-described tasks in a code base the programmer actually understands. Qwen3.6 27B and DeepSeek V4 Flash are both great little workhorses.
There's also GLM 5.2, which is kind of like "store brand Opus", and which might be considered a "near-frontier" model. I don't have as much experience with it.
FWIW Fable is insanely expensive for the task you just described, so much so that I don't think it's practical for that. Its practical use is as a dev lead / architect / project manager model, doing planning and writing detailed feature specs and code reviews while Opus/Codex/Gemini does the actual coding.
reply