Back to leaderboard

qwen3:4b

OLLAMA GGUF

qwen3 · 4.0B · Q4_K_M

Not Rec.

Jun 26, 2026 · Cortex-A78AE

Kimi-48B:48B

OLLAMA GGUF

kimi-linear · 49.1B · unknown

Excellent

Aug 24, 2026 · Intel Core™ Ultra X7 358H

Global Score
32 vs 86
Hardware Fit
98 vs 78
Quality Score
4 vs 89

Hardware

qwen3:4b Kimi-48B:48B
MachineNVIDIA NVIDIA Jetson Orin Nano Engineering Reference Developer Kit SuperGMKtec NucBox EVO-T2S
CPUCortex-A78AEIntel Core™ Ultra X7 358H
Cores616
RAM7 GB64 GB
GPUIntegrated / UnknownIntel(R) Arc(TM) B390 GPU
OSUbuntu 22.04.5 LTSMicrosoft Windows 11 Pro 10.0.26200
Archarm64x64
Power Modebalancedbalanced

Performance

qwen3:4b Kimi-48B:48B
Tokens/sec17.923.1
First chunk950 ms752 ms
TTFT950 ms752 ms
Load time9.7 s22.2 s
Memory usage3.0 GB25.5 GB
Memory %40%40%

HW Fit Score Breakdown

qwen3:4b

Speed
50/50
TTFT
20/20
Memory
28/30

Kimi-48B:48B

Speed
31/50
TTFT
20/20
Memory
27/30

Quality

qwen3:4b

Reasoning
4/20
Coding
0/20
Instruction
0/20
Structured
0/15
Math
0/15
Multilingual
0/10
Reasoning: Poor Coding: Poor Instruction Following: Poor Structured Output: Poor Math: Poor Multilingual: Poor

Kimi-48B:48B

Reasoning
18/20
Coding
19/20
Instruction
15/20
Structured
15/15
Math
13/15
Multilingual
9/10
Reasoning: Strong Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Strong Multilingual: Strong

Run yours and compare

$ npm install -g metrillm@latest
$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest