Kosmos GGUF models
127 GGUF models in the Kosmos family, quantized by 8 uploaders. Sizes run from 8.03B to 8.03B parameters. The longest trained context is 128K tokens.
About the Kosmos GGUF models
Kosmos is represented on ggufmodels.com by 127 GGUF models, quantized by 8 uploaders on Hugging Face. The most downloaded are Kosmos EVAA v9 8B i1, Kosmos EVAA 8B i1 and Kosmos EVAA TSN light 8B. Parameter counts run from 8.03B to 8.03B, and the longest trained context window is 128K tokens.
A Q4_K_M file of a Kosmos model is between 4.9 to 4.9 GB, which is the quant most people start with. Most files come from jaspionjader (63), mradermacher (51) and MaziyarPanahi (6). The quants you will meet most often are Q5_K_M (121), Q6_K (60), Q5_K_S (59) and Q4_K_M (55), where the number is how many models offer that quant.
Frequently asked questions
Which Kosmos GGUF should I download?
The most downloaded Kosmos GGUF here is Kosmos EVAA v9 8B i1. Pick the size that fits your GPU first, then choose a quant: Q4_K_M is the usual starting point, and Q5_K_M or Q6_K give more quality if you have memory to spare. The VRAM pages below list what fits each GPU.
How much VRAM does a Kosmos model need?
Q4_K_M files of Kosmos models on this site range from 4.9 to 4.9 GB. Add the KV cache for your context length and about 0.5 GB for the runtime, which usually means 1.5 to 3 GB on top of the file at an 8K context.
Where do Kosmos GGUF files come from?
The most common quantizers here are jaspionjader, mradermacher and MaziyarPanahi. They convert the original model and publish quantized files on Hugging Face. ggufmodels.com does not host files; every download button opens Hugging Face.