IQ2_XXS GGUF models

Importance-matrix 2-bit quant. Extreme compression with a large quality cost. 26647 models have a IQ2_XXS file. Sizes run from 2M to 2780B parameters. The longest trained context is 10240K tokens.

Most downloadedNewestLargestLongest contextSmallest
OL
Olmo-3.1-32B-Instruct
OLMo · unsloth
Params32.2B
Context64K
Q4_K_M19.5 GB
GGUF
QW
Qwen3.5 35B A3B Uncensored Heretic
Qwen · mradermacher
Params34.7B
Context256K
Q4_K_M21.2 GB
GGUF
IC
IceBlendedLatteRP-7b
Iceblendedlatterp · mradermacher
Params7.24B
Context32K
Q4_K_M4.4 GB
GGUF
CO
Concept 7b V1 Full
Concept · mradermacher
Params7.62B
Context128K
Q4_K_M4.7 GB
GGUF
SO
Sorjin1.1 7B Instruct
Sorjin · mradermacher
Params7.62B
Context32K
Q4_K_M4.7 GB
GGUF
QW
Qwen3.5 9B Unredacted MAX
Qwen · mradermacher
Params8.95B
Context256K
Q4_K_M5.6 GB
Reasoning
QW
Qwen3 30B A3B Thinking 2507 FT
Qwen · mradermacher
Params30.5B
Context256K
Q4_K_M18.6 GB
Reasoning
QW
Qwen3.5 35B A3B Base
Qwen · mradermacher
Params34.7B
Context256K
Q4_K_M21.2 GB
GGUF
HO
K2 Horizon 0.9B
Horizon · NANI-Nithin
Params1.08B
Context128K
Q4_K_M0.7 GB
GGUF
LL
Hermes 3 Llama 3.1 70B
Llama · mradermacher
Params70.6B
Context128K
Q4_K_M42.5 GB
Tool use
GE
Gemma 4 31B It Claude Opus Distill v2 Heretic
Gemma · mradermacher
Params30.7B
Context256K
Q4_K_M18.7 GB
Reasoning
EI
Einstein-v4-7B
Einstein · mradermacher
Params7.24B
Context32K
Q4_K_M4.4 GB
CodingTool use
CI
Cirque 12B
Cirque · mradermacher
Params12.2B
Context128K
Q4_K_M7.5 GB
GGUF
LL
Meta-Llama-3-8B-Instruct
Llama · qwp4w3hyb
Params8B
Context8K
Q4_K_M4.9 GB
GGUF
DE
DeepSeek R1 Distill Llama 70B Uncensored
DeepSeek · mradermacher
Params70.6B
Context128K
Q4_K_M42.5 GB
Reasoning
PH
calme-2.1-phi3.5-4b
Phi · mradermacher
Params3.82B
Context128K
Q4_K_M2.4 GB
GGUF
LL
Lebanese Llama 3.1 8b
Llama · mradermacher
Params8B
Context128K
Q4_K_M4.9 GB
GGUF
NE
NeonLLM-Base
Neonllm · mradermacher
Params7.24B
Context8K
Q4_K_M4.4 GB
Tool use
NA
Nano.Imp Creative.Writing Heretic 3.2 1B
Nano · mradermacher
Params1.5B
Context128K
Q4_K_M1.0 GB
Roleplay
HA
Harbinger 24B
Harbinger · LatitudeGames
Params23.6B
Context128K
Q4_K_M14.3 GB
Roleplay
DE
WhiteRabbitNeo V3 7B i1
Deephat · mradermacher
Params7.62B
Context32K
Q4_K_M4.7 GB
Coding
NA
Nanbeige4.3 33B Base
Nanbeige · mradermacher
Params33.1B
Context128K
Q4_K_M20.1 GB
GGUF
LL
MultiPL-T-CodeLlama_70b
Llama · mradermacher
Params69B
Context16K
Q4_K_M41.4 GB
Coding
PH
Iceball Phi 3.5 mini instruct shareGPT v1.0 i1
Phi · mradermacher
Params3.82B
Context128K
Q4_K_M2.3 GB
GGUF
PH
Phi4 Abliterated
Phi · mradermacher
Params14.7B
Context16K
Q4_K_M9.1 GB
GGUF
ME
G4 MeroMero 26B A4B
Meromero · mradermacher
Params25.2B
Context256K
Q4_K_M16.8 GB
Roleplay
YI
Yi-34B-200K
Yi · mradermacher
Params34.4B
Context195K
Q4_K_M20.7 GB
GGUF
QW
Git Commit Message Splitter Qwen3 8B
Qwen · mradermacher
Params8.19B
Context40K
Q4_K_M5.0 GB
GGUF
LL
calme-2.1-llama3.1-70b
Llama · mradermacher
Params70.6B
Context128K
Q4_K_M42.5 GB
GGUF
QW
Qwen3 DND Jan v1 256k Ctx Brainstorm40x 8B
Qwen · mradermacher
Params7.66B
Context256K
Q4_K_M4.7 GB
CodingReasoning

Other quants

Keep browsing