CAPTAIN ERIS NEBULA (CHIMERA) / I-MATRIX / 12B / I-QUANT

Evolution of a personal favorite model: Captain Eris Violet, by the same creator, Nitral-AI. This model demonstrates good storytelling compared to 21b and 27b MoE models, and 18b models. This also serves as the last open-source model to be made available by the creator for the forseeable future as of November 2025. To stuff as many parameters in as little VRAM as possible, weighted I-quants will be listed.

Note that I-quants forfeit some token generation speed relative to K-quants in exchange for storage efficiency. Any model smaller than the extra small 4-bit I-quant will fit on 8GB GPUs. All models will fit well on 16GB GPUs, including the ‘full’ 8-bit version. Weighted quants differ from static quants with the ‘importance’ of each weight with the lowering of the bit depth taken into consideration; as the 8-bit model is the ‘full’ model, weighted quants are not required, and the uploaded Q8_0 model is a static quant. These models were taken from GGUF formats from Huggingface.

Original model (Nitral-AI):

GGUF weighted quantizations (mradermacher):

GGUF static quantizations (mradermacher):