GGUF models by kolosal

41 GGUF models quantized by kolosal on Hugging Face. Sizes run from 494M to 32.8B parameters. The longest trained context is 128K tokens.

Most downloadedNewestLargestLongest contextSmallest
QW
Qwen2.5 Coder 14B Instruct AWQ
Qwen · kolosal
Params14.8B
Context128K
Q4_K_M18.0 GB
CodingTool use
QW
Qwen2.5 Coder 3B Instruct AWQ
Qwen · kolosal
Params3.4B
Context32K
Q4_K_M2.1 GB
CodingTool use
QW
Qwen2.5 Coder 1.5B Instruct GGUF
Qwen · kolosal
Params1.54B
Context32K
Q4_K_M1.1 GB
CodingTool use
QW
Qwen2.5 Coder 7B Instruct GGUF
Qwen · kolosal
Params7.62B
Context128K
Q4_K_M9.4 GB
CodingTool use
QW
Qwen2.5 Coder 0.5B Instruct AWQ
Qwen · kolosal
Params630M
Context32K
Q4_K_M0.5 GB
CodingTool use
GE
Google_Gemma 3 4b It
Gemma · kolosal
Params3.88B
Context128K
Q4_K_M2.5 GB
GGUF
DE
Deepseek R1 Qwen 1.5B
DeepSeek · kolosal
Params1.78B
Context128K
Q4_K_M1.1 GB
ReasoningTool use
LL
llama 3 8b sahabat ai
Llama · kolosal
Params8.03B
Context8K
Q4_K_M4.9 GB
GGUF
LL
Llama 3.2 3B Instruct
Llama · kolosal
Params3.21B
Context128K
Q4_K_M2.0 GB
Tool use
QW
Qwen2.5 3B Instruct
Qwen · kolosal
Params3.09B
Context32K
Q4_K_M1.9 GB
Tool use
DE
Deepseek R1 Qwen 14B
DeepSeek · kolosal
Params14.8B
Context128K
Q4_K_M9.0 GB
ReasoningTool use
QW
Qwen2.5 7B Instruct
Qwen · kolosal
Params7.62B
Context32K
Q4_K_M4.7 GB
Tool use
MI
mistralai_Ministral 8B Instruct 2410
Mistral · kolosal
Params8.02B
Context32K
Q4_K_M4.9 GB
Tool use
GE
Google_Gemma 3 12b It
Gemma · kolosal
Params11.8B
Context128K
Q4_K_M7.3 GB
GGUF
BA
bahasa ai 4b
Bahasa · kolosal
Params3.95B
Context32K
Q4_K_M2.5 GB
GGUF
LL
Meta Llama 3.1 8B Instruct
Llama · kolosal
Params8.03B
Context128K
Q4_K_M4.9 GB
GGUF
QW
Qwen3 Embedding 4B Bf16
Qwen · kolosal
Params4.02B
Context40K
Q4_K_M2.5 GB
EmbeddingTool use
PH
phi 4
Phi · kolosal
Params14.7B
Context16K
Q4_K_M8.9 GB
GGUF
QW
QwQ 32B
Qwen · kolosal
Params32.8B
Context128K
Q4_K_M19.9 GB
ReasoningTool use
QW
Qwen3-0.6B
Qwen · kolosal
Params596M
Context40K
Smallest0.4 GB
ReasoningTool use
GP
Openai_Gpt Oss 20b
gpt-oss · kolosal
Params20.9B
Context128K
Smallest12.1 GB
Tool use
GE
Gemma 2 2b It
Gemma · kolosal
Params2.61B
Context8K
Q4_K_M1.7 GB
GGUF
QW
Qwen3-1.7B
Qwen · kolosal
Params1.72B
Context40K
Smallest1.1 GB
ReasoningTool use
QW
Qwen3-30B-A3B
Qwen · kolosal
Params30.5B
Context40K
Smallest17.7 GB
ReasoningTool use
MI
mistral 7b v0.3
Mistral · kolosal
Params7.25B
Context32K
Q4_K_M4.4 GB
Tool use
QW
Qwen3-8B-128K
Qwen · kolosal
Params8.19B
Context128K
Smallest5.2 GB
ReasoningTool use
QW
Qwen3-4B-128K
Qwen · kolosal
Params4.02B
Context128K
Smallest2.5 GB
ReasoningTool use
GE
gemma 2 9b sahabat ai
Gemma · kolosal
Params9.24B
Context8K
Q4_K_M5.8 GB
GGUF
QW
Qwen2.5 1.5B Instruct
Qwen · kolosal
Params1.54B
Context32K
Q4_K_M1.0 GB
Tool use
QW
Qwen2.5 0.5B Instruct
Qwen · kolosal
Params494M
Context32K
Q4_K_M0.4 GB
Tool use

About kolosal on Hugging Face

kolosal has published 41 GGUF models that ggufmodels.com tracks, across families such as Qwen, Gemma and Llama. The most downloaded are Qwen2.5 Coder 14B Instruct AWQ, Qwen2.5 Coder 3B Instruct AWQ and Qwen2.5 Coder 1.5B Instruct GGUF.

The quants kolosal publishes most often are Q8_0 (34), Q4_K_M (31), F16 (23) and BF16 (9). Their Q4_K_M files range from 0.4 to 19.9 GB. Parameter counts run from 494M to 32.8B, and the longest trained context window is 128K tokens.

Frequently asked questions

What GGUF models has kolosal published?

41 models are listed here, mainly from the Qwen, Gemma and Llama families. The most downloaded is Qwen2.5 Coder 14B Instruct AWQ.

Which quants does kolosal publish?

The most common are Q8_0 (34 models), Q4_K_M (31 models), F16 (23 models) and BF16 (9 models).

Keep browsing