List your company

Inference Efficiency on On-device Samsung Galaxy S25 (1k, 2k, 4k context)

0.84Prefill TTFT (1k)LFM2 350M
0.68841.71172.7353.7583Mar 16, 2026

Evaluation results

MethodLinks
LFM2 350MParameter Count=350M, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.030.842.184.52147.5496.963.6
MobileLLM-Flash 350MParameter Count=350M, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.030.912.785.33165.56112.5895.55
Gemma3 270MParameter Count=270M, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.030.993.135.64123.7182.8552.4
MobileLLM-Flash 650MParameter Count=650M, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.031.623.348.4896.6485.3560.74
LFM2 700MParameter Count=700M, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.032.056.018.2486.353.5744.36
MobileLLM-Flash 1.4BParameter Count=1.4B, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.033.49.0816.560.5242.6534.01
LFM2 1.2BParameter Count=1.2B, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.033.468.4116.8861.1442.1529.14
Gemma3 1BParameter Count=1B, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.033.559.2918.8658.5843.1136.29
Llama3.2 1BParameter Count=1B, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.034.5115.0926.9539.0118.1314.34
Qwen3 0.6BParameter Count=0.6B, Hardware=Samsung Galaxy S25, CPU Threads=4, Quantization=4-bit weight, 8-bit activation2026.034.6311.5925.444.5628.6718.48