Zephyr GGUF models
250 GGUF models in the Zephyr family, quantized by 70 uploaders. Sizes run from 7M to 141B parameters. The longest trained context is 195K tokens.
About the Zephyr GGUF models
Zephyr is represented on ggufmodels.com by 250 GGUF models, quantized by 70 uploaders on Hugging Face. The most downloaded are huggingfaceh4_zephyr-7b-beta, zephyr orpo 141b A35b v0.1 GGUF smashed and Zephyr 7B Beta Harmless. Parameter counts run from 7M to 141B, and the longest trained context window is 195K tokens.
A Q4_K_M file of a Zephyr model is between 0.1 to 85.6 GB, which is the quant most people start with. Most files come from mradermacher (77), RichardErkhov (42) and tensorblock (23). The quants you will meet most often are Q2_K (181), Q4_K_M (178), Q3_K_M (163) and Q5_K_M (162), where the number is how many models offer that quant.
Frequently asked questions
Which Zephyr GGUF should I download?
The most downloaded Zephyr GGUF here is huggingfaceh4_zephyr-7b-beta. 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 Zephyr model need?
Q4_K_M files of Zephyr models on this site range from 0.1 to 85.6 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 Zephyr GGUF files come from?
The most common quantizers here are mradermacher, RichardErkhov and tensorblock. They convert the original model and publish quantized files on Hugging Face. ggufmodels.com does not host files; every download button opens Hugging Face.