The quantized models run on machines with 32GB+ of RAM.
Unsloth AI (@UnslothAI) released new Qwen3.8-Flash-Next GGUFs, claiming 15% higher accuracy.
The company says its Dynamic V3.1 quantization beats others by more than 15% on Terminal Bench, DeepSWE and other benchmarks, with 256K context. The models run on setups with 32GB or more of RAM.
The release drew about 738 likes.
A year in the making.
@min2hound9.6k
Support for M4 through M6 is on the way.
@dhh5k
The open-source PS3 emulator reaches a milestone.
@Pirat_Nation2.7k
The company pitches it as proof that AI-written code at scale doesn't have to be slop.
@axiommathai2.2k
A 19,000-character Japanese guide went viral.
@Yacamochi_db2.1k
It would also become the 13th most-starred repo of all time.
@mattpocockuk1.5kOne email each morning with the AI, crypto and tech posts everyone is talking about, explained in plain English.
