IQ4_XS GGUF models

Importance-matrix 4-bit quant, extra small. Small files with good quality for the size. 97368 models have a IQ4_XS file. Sizes run from 1M to 2450B parameters. The longest trained context is 102400K tokens.

Most downloadedNewestLargestLongest contextSmallest
PH
Phi 4 Mini Instruct
Phi · bartowski
Params3.84B
Context128K
Q4_K_M2.5 GB
GGUF
GE
Gemma-3-12B-It-Qat
Gemma · unsloth
Params11.8B
Context128K
Q4_K_M7.3 GB
Vision
QW
Qwen3.8 Flash Next
Qwen · bartowski
Params177B
Context256K
Q4_K_M119.6 GB
Vision
QW
Qwen3.8 27B Uncensored
Qwen · orcarouter
Params27.3B
Context256K
Q4_K_M16.8 GB
ReasoningTool useVision
GE
Gemma 4 26B A4B It Ultra Uncensored Heretic
Gemma · mradermacher
Params25.2B
Context256K
Q4_K_M16.8 GB
GGUF
QW
Swift 1.5 Qwen3.8 27B Uncensored Dynamic MTP
Qwen · ajgazin
Params28B
Context256K
Smallest8.8 GB
Vision
GE
gemma 4 12b heretic abliterated
Gemma · culturerevolt
Params11.9B
Context256K
Q4_K_M7.4 GB
Vision
QW
Qwen3-Vl-2B-Instruct
Qwen · unsloth
Params1.72B
Context256K
Q4_K_M1.1 GB
Vision
QW
Qwen3.8 27B Humanlike Chat
Qwen · LessThanThreeAI
Params26.9B
Context256K
Q4_K_M16.6 GB
ReasoningRoleplayTool use
GE
Gemma-3-270M-It
Gemma · unsloth
Params268M
Context32K
Q4_K_M0.3 GB
GGUF
QW
Huihui Qwen3.6 35B A3B Abliterated
Qwen · mradermacher
Params34.7B
Context256K
Q4_K_M21.2 GB
Vision
GE
Gemma4-26B-A4B-Uncensored-HauhauCS-Balanced
Gemma · HauhauCS
Params25.2B
Context256K
Q4_K_M16.8 GB
Vision
QW
Huihui Qwen3 Coder 30B A3B Instruct Abliterated
Qwen · mradermacher
Params30.5B
Context256K
Q4_K_M18.6 GB
Coding
HE
Hermes 4 14B
Hermes · bartowski
Params14.8B
Context40K
Q4_K_M9.0 GB
Tool use
MI
Devstral-Small-2-24B-Instruct-2512
Mistral · unsloth
Params23.6B
Context384K
Q4_K_M14.3 GB
CodingVision
QW
Qwen3.5-9B
Qwen · unsloth
Params9.2B
Context256K
Q4_K_M5.9 GB
Vision
NA
Nanbeige4.3 33B
Nanbeige · bartowski
Params33.1B
Context512K
Q4_K_M20.8 GB
ReasoningTool use
QW
Huihui Qwen3.8 Flash Next Abliterated
Qwen · mradermacher
Params177B
Context256K
Smallest80.4 GB
Vision
GR
Granite 4.0 1B BE
Granite · taronaeo
Params1.63B
Context—
Q4_K_M1.0 GB
GGUF
QW
Qwen2.5 1.5B Instruct
Qwen · bartowski
Params1.54B
Context32K
Q4_K_M1.0 GB
GGUF
LL
Meta Llama 3.1 8B Instruct
Llama · lmstudio-community
Params8B
Context128K
Q4_K_M4.9 GB
GGUF
QW
Qwen2.5 3B Instruct
Qwen · bartowski
Params3.09B
Context32K
Q4_K_M1.9 GB
GGUF
DE
DeepSeek R1 Distill Qwen 32B
DeepSeek · bartowski
Params32.8B
Context128K
Q4_K_M19.9 GB
Reasoning
TO
Topxtral-4x7B-v0.1
Topxtral · RichardErkhov
Params18.5B
Context32K
Q4_K_M11.2 GB
GGUF
QW
Qwen3 4B Instruct 2507
Qwen · bartowski
Params4B
Context256K
Q4_K_M2.5 GB
GGUF
QW
Qwen3 30B A3B Instruct 2507
Qwen · bartowski
Params30.5B
Context256K
Q4_K_M18.6 GB
GGUF
QW
Qwen3.6 40B Claude 4.6 Opus Deckard Heretic Uncensored Thinking NEO CODE Di IMatrix MAX
Qwen · DavidAU
Params39.1B
Context256K
Q4_K_M24.3 GB
CodingReasoningRoleplay
QW
Qwen3-Next-80B-A3B-Instruct
Qwen · unsloth
Params80B
Context256K
Q4_K_M48.5 GB
GGUF
DE
DeepSeek R1 Distill Qwen 14B
DeepSeek · bartowski
Params14.8B
Context128K
Q4_K_M9.0 GB
Reasoning
MI
Mistral-Small-3.2-24B-Instruct-2506
Mistral · unsloth
Params23.6B
Context128K
Q4_K_M14.3 GB
Vision

Other quants

Keep browsing