Viable GGUFs without excessive loss at Q4 and better for people who have either 256GB or 512GB inference systems.
We're seeing a real flurry of 'very capable' open weight models release in the last 3-4 days, including Qwen 3.8-Flash-Next which fits on a 256GB system in Q8.
Discussion (1 Comments)Read Original on HackerNews
We're seeing a real flurry of 'very capable' open weight models release in the last 3-4 days, including Qwen 3.8-Flash-Next which fits on a 256GB system in Q8.
https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF