baidu/ernie-4.5-21b-a3b
LM-STUDIO GGUFernie4_5-moe · 21B · Q4_K_M
Nov 1, 2026 · AMD RYZEN AI MAX+ 395
nemotron-3-nano
LM-STUDIO GGUFMar 6, 2026 · Apple M4 Pro
Global Score
86 vs 93
Hardware Fit
93 vs 100
Quality Score
83 vs 90
Hardware
baidu/ernie-4.5-21… nemotron-3-nano
MachineGMKtec EVO-X3Mac mini
CPUAMD RYZEN AI MAX+ 395Apple M4 Pro
Cores3214
RAM120 GB64 GB
GPUAMD Radeon(TM) 8060S GraphicsApple M4 Pro
OSMicrosoft Windows 11 Pro 10.0.26300macOS 15.7.4
Archx64arm64
Power Modebalancedbalanced
Performance
baidu/ernie-4.5-21… nemotron-3-nano
Tokens/sec46.393.3
First chunk33 ms257 ms
TTFT383 ms257 ms
Load time49.7 sN/A
Memory usage32.5 GB0.1 GB
Memory %27%0%
HW Fit Score Breakdown
baidu/ernie-4.5-21b-a3b
Speed
43/50
TTFT
20/20
Memory
30/30
nemotron-3-nano
Speed
50/50
TTFT
20/20
Memory
30/30
Quality
baidu/ernie-4.5-21b-a3b
Reasoning
16/20
Coding
17/20
Instruction
15/20
Structured
14/15
Math
12/15
Multilingual
9/10
Reasoning: Strong Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Strong Multilingual: Strong
nemotron-3-nano
Reasoning
19/20
Coding
18/20
Instruction
14/20
Structured
15/15
Math
14/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Adequate Structured Output: Strong Math: Strong Multilingual: Strong
Run yours and compare
$
npm install -g metrillm@latest$
metrillmRequires Node 20+ and Ollama or LM Studio running
Or run without installing: npx metrillm@latest