Nex-N2.5 Mini uncensored model drops for local inference
The Nex-N2.5 Mini Uncensored model, trained on the Qwen3.5-35B-A3B architecture, is now available. This release caters to the community's demand for uncensored local models and is noted for being highly practical to run, offered in a compact IQ4_XS quantization format.
The release highlights the persistent community drive to create and run uncensored, open-weights models locally without safety guardrails.
- –Built on top of the Qwen3.5-35B-A3B foundation.
- –Packaged in IQ4_XS quantization, making it accessible for consumer hardware.
- –Reflects ongoing trends of removing alignment filters for personal, local AI instances.
DISCOVERED
1h ago
2026-09-11
PUBLISHED
1h ago
2026-09-11
RELEVANCE
AUTHOR
Oluwaphilemon1