Celeste GGUF models
73 GGUF models in the Celeste family, quantized by 21 uploaders. Sizes run from 8B to 70.6B parameters. The longest trained context is 1000K tokens.
About the Celeste GGUF models
Celeste is represented on ggufmodels.com by 73 GGUF models, quantized by 21 uploaders on Hugging Face. The most downloaded are MN 12B Celeste V1.9, MN 12B Celeste V1.9 and L3.1 8B Celeste V1.5. Parameter counts run from 8B to 70.6B, and the longest trained context window is 1000K tokens.
A Q4_K_M file of a Celeste model is between 4.9 to 42.5 GB, which is the quant most people start with. Most files come from mradermacher (20), matrixportalx (8) and aaronday3 (6). The quants you will meet most often are Q6_K (50), Q4_K_M (48), Q5_K_M (42) and Q4_K_S (42), where the number is how many models offer that quant.
Frequently asked questions
Which Celeste GGUF should I download?
The most downloaded Celeste GGUF here is MN 12B Celeste V1.9. 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 Celeste model need?
Q4_K_M files of Celeste models on this site range from 4.9 to 42.5 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 Celeste GGUF files come from?
The most common quantizers here are mradermacher, matrixportalx and aaronday3. They convert the original model and publish quantized files on Hugging Face. ggufmodels.com does not host files; every download button opens Hugging Face.