GGUF models by elichen-skymizer

91 GGUF models quantized by elichen-skymizer on Hugging Face. Sizes run from 596M to 30.5B parameters. The longest trained context is 1024K tokens.

Most downloadedNewestLargestLongest contextSmallest
QW
Qwen3 4B Thinking 2507
Qwen · elichen-skymizer
Params4.02B
Context256K
Smallest2.1 GB
ReasoningTool use
QW
Qwen3 4B Thinking 2507
Qwen · elichen-skymizer
Params4.02B
Context256K
Q4_K_M2.5 GB
ReasoningTool use
QW
Qwen2.5 VL 3B Instruct
Qwen · elichen-skymizer
Params3.09B
Context125K
Q4_K_M1.9 GB
Vision
GE
Gemma 3 1b Pt
Gemma · elichen-skymizer
Params1B
Context32K
Q4_K_M0.8 GB
GGUF
QW
Qwen3 VL 4B Instruct
Qwen · elichen-skymizer
Params4.02B
Context256K
Q4_K_M2.5 GB
Tool useVision
QW
Qwen3 30B A3B Instruct 2507
Qwen · elichen-skymizer
Params30.5B
Context256K
Smallest14.7 GB
ReasoningTool use
QW
Qwen2.5 VL 3B Instruct
Qwen · elichen-skymizer
Params3.09B
Context125K
Smallest1.6 GB
Vision
QW
Qwen2.5 VL 7B Instruct
Qwen · elichen-skymizer
Params7.62B
Context125K
Smallest3.8 GB
Vision
YU
yuko q3 models
Yuko · elichen-skymizer
Params4.02B
Context256K
Smallest2.1 GB
ReasoningTool use
MM
mmlu eval models
Mmlu · elichen-skymizer
Params6.74B
Context4K
Smallest13.5 GB
GGUF
DE
DeepSeek R1 Distill Qwen 1.5B
DeepSeek · elichen-skymizer
Params1.78B
Context128K
Q4_K_M1.1 GB
ReasoningTool use
QW
Qwen3 VL 30B A3B Thinking
Qwen · elichen-skymizer
Params30.5B
Context256K
Smallest14.7 GB
ReasoningTool useVision
QW
Qwen3 4B Thinking 2507
Qwen · elichen-skymizer
Params4.02B
Context256K
Smallest3.3 GB
ReasoningTool use
QW
Qwen3 VL 30B A3B Instruct
Qwen · elichen-skymizer
Params30.5B
Context256K
Q4_K_M18.6 GB
Tool useVision
QW
Qwen2.5 VL 7B Instruct
Qwen · elichen-skymizer
Params7.62B
Context125K
Q4_K_M4.7 GB
Vision
QW
Qwen3 30B A3B Thinking 2507
Qwen · elichen-skymizer
Params30.5B
Context256K
Smallest14.7 GB
ReasoningTool use
GR
Granite 3.1 1b A400M Base
Granite · elichen-skymizer
Params1.33B
Context128K
Smallest0.7 GB
GGUF
QW
Qwen3 30B A3B Instruct 2507
Qwen · elichen-skymizer
Params30.5B
Context256K
Smallest32.5 GB
ReasoningTool use
LL
Llama 3.2 3B Instruct
Llama · elichen-skymizer
Params3.21B
Context128K
Smallest2.3 GB
Tool use
GE
Gemma 3 1b Pt
Gemma · elichen-skymizer
Params1B
Context32K
Smallest1.0 GB
GGUF
GE
Gemma 3 1b Pt
Gemma · elichen-skymizer
Params1B
Context32K
Smallest0.7 GB
GGUF
GR
Granite 3.1 1b A400M Base
Granite · elichen-skymizer
Params1.33B
Context128K
Smallest1.1 GB
GGUF
QW
Qwen3 VL 30B A3B Thinking
Qwen · elichen-skymizer
Params30.5B
Context256K
Q4_K_M18.6 GB
ReasoningTool useVision
QW
Qwen3 30B A3B Instruct 2507
Qwen · elichen-skymizer
Params30.5B
Context256K
Q4_K_M18.6 GB
ReasoningTool use
LL
Llama 3.2 3B
Llama · elichen-skymizer
Params3.21B
Context128K
Smallest3.4 GB
GGUF
LL
Llama 3.2 3B Instruct
Llama · elichen-skymizer
Params3.21B
Context128K
Smallest2.6 GB
Tool use
DE
DeepSeek R1 Distill Qwen 1.5B
DeepSeek · elichen-skymizer
Params1.78B
Context128K
Smallest1.1 GB
ReasoningTool use
QW
Qwen3 VL 30B A3B Instruct
Qwen · elichen-skymizer
Params30.5B
Context256K
Smallest14.7 GB
Tool useVision
QW
Qwen3 30B A3B Thinking 2507
Qwen · elichen-skymizer
Params30.5B
Context256K
Smallest32.5 GB
ReasoningTool use
QW
Qwen3 VL 4B Instruct
Qwen · elichen-skymizer
Params4.02B
Context256K
Smallest2.1 GB
Tool useVision

About elichen-skymizer on Hugging Face

elichen-skymizer has published 91 GGUF models that ggufmodels.com tracks, across families such as Qwen, Llama and DeepSeek. The most downloaded are Qwen3 4B Thinking 2507, Qwen3 4B Thinking 2507 and Qwen2.5 VL 3B Instruct.

The quants elichen-skymizer publishes most often are Q3_K_M (22), Q4_K_M (21), Q6_K (16) and Q5_K_M (16). Their Q4_K_M files range from 0.4 to 18.6 GB. Parameter counts run from 596M to 30.5B, and the longest trained context window is 1024K tokens.

Frequently asked questions

What GGUF models has elichen-skymizer published?

91 models are listed here, mainly from the Qwen, Llama and DeepSeek families. The most downloaded is Qwen3 4B Thinking 2507.

Which quants does elichen-skymizer publish?

The most common are Q3_K_M (22 models), Q4_K_M (21 models), Q6_K (16 models) and Q5_K_M (16 models).

Keep browsing