PocketDoc/Dans-SakuraKaze-V1.0.0-12b

Updated Feb 14 • 1.03k • 19

Note Feels like the best 12B RP yet

Nitral-AI/Captain_Eris_Noctis-12B-v0.420

Text Generation • Updated Jan 16 • 130 • 20

Note This was the first Nemo model I really liked

saturated-labs/T-Rex-mini

Updated 8 days ago • 63 • 17

Note Surprisingly good for its size, follows instructions well enough, a good start if your VRAM is limited

SicariusSicariiStuff/Wingless_Imp_8B

Updated Mar 9 • 17 • 10

Note High IFeval for an 8B model that is not too censored: 74.30 - nuff said

stduhpf/google-gemma-3-12b-it-qat-q4_0-gguf-small

Updated 3 days ago • 5.02k • 29

Note If you want to try out the latest Gemma model - don't use the abliterated versions. A well-constructed system promt is sufficient for this model. The general advice is not to use a quantized KV cache. Keep it to 8-bit at most, anything less will mess up your context very badly