BenchClub
Overview
Leaderboard
Models
Labs
Creators
Test sets
Search models
Sign in
Formats
Quantized builds on Hugging Face by file format, counted Nov 18, 2026.
By size
Labs
Releases
Families
Licenses
Most built on
Formats
All models
GGUF
12,111 builds
233
models
llama.cpp, Ollama, LM Studio
MLX
3,434 builds
188
models
Apple silicon: mlx-lm, LM Studio
AWQ
574 builds
129
models
vLLM, SGLang
GPTQ
821 builds
130
models
vLLM, SGLang
All sizes
≤3B
3–9B
9–20B
20–40B
40–80B
80–300B
>300B
Top build from
Anyone
mlx-community
85
lmstudio-community
36
pipenetwork
3
baa-ai
3
Jundot
2
TensorFold
2
LiquidAI
2
inferencerlabs
2
#
Model
Size
Score
GGUF
MLX
AWQ
GPTQ
Most downloaded MLX
1
L
Llama-4-Maverick-17B-128E-Instruct
Meta Llama
402B
1287
8
3
1
—
Llama-4-Maverick-17B-128E-Instruct-RAM-170GB-MLX
baa-ai · 200 downloads
2
D
DeepSeek-V3.2
DeepSeek
685B
1425
6
2
1
—
DeepSeek-V3.2-RAM-350GB-MLX
baa-ai · 402 downloads
3
L
Llama-3.1-70B-Instruct
Meta Llama
70.6B
1261
37
1
2
2
Llama-3.1-70B-Instruct-SWAN-5bit-MLX
baa-ai · 248 downloads