Local model picks by VRAM tier, from 4GB to 384GB
The author recommends clef-flash for the 8–16GB tier, saying it works like Jev and suits browser use, automation and high-volume processing.
Best models for your hardware October —— 4GB to 8GB VRAM —— Smoll but mighty https://huggingface.co/LiquidAI/LFM2.5-2.6B —— 8GB to 16GB VRAM —— 1. Underdog Saluki https://huggingface.co/ConwayResearch/Underdog-Saluki-27B-1.0 this is the closest you’ll get to frontier on this memory tier It’s a custom post train of Qwen Or if you want a classifier 2. Clef-flash: https://huggingface.co/Cloudflare/clef-flash this works like JEV great for browser use, automation and high volume processing —— 16GB to 32GB VRAM —— 1. Qwen3.8-27b The greatest model of all time, qwen3.8-27B https://huggingface.co/Qwen/Qwen3.8-27B This model can code, edit videos, hack, do 3d work, admin, browser use etc.. 2. Lfm: if you’re on slower hardware I highly recommend https://huggingface.co/LiquidAI/LFM2.5-24B-A2B incredible performance —— 32GB to 96GB VRAM —— If you have 64gb or more of ddr4/ddr5 memory then you can run https://huggingface.co/Qwen/Qwen3.8-Flash-Next with offloading —— 96GB to 192GB of VRAM —— You can go with either the model above https://huggingface.co/Qwen/Qwen3.8-Flash-Next Or GLM-5.3-Flash https://huggingface.co/zai-org/GLM-5.3-flash —— 192GB to 384GB ——- You can either go with GLM-5.3-Flash or deepseek-v4.1-flash