Neutrino-8B
by Fermion Research · Model · Language & code models
8B model with sub-2-bit weights, shipped as a 2.56 GB download container with Apache-2.0 weights, based on Qwen3-8B
AccessOpen weightsNeeds~10 GBDownload4.1 GBSize8B parametersLicenseApache-2.0
Evidence
- Aug 9Answered a dinosaur trick question correctly but looped, circling the same point repeatedly before answering; judged weaker than Bonsai on this
- Aug 9Served on a 48 GB GPU via llama.cpp fork, using just under 10 GB of VRAM; loaded fast
Cost to run
Alternatives
Made from