modelgrep

Google: Gemma 3 4B vs Meta: Llama 3.2 1B Instruct

Google: Gemma 3 4B wins on more metrics (3 of 4), but the right pick depends on what you optimize for — see the breakdown below.

MetricGoogle: Gemma 3 4BMeta: Llama 3.2 1B Instruct
Intelligence Index6.36.3
Coding Index2.90.6
GPQA Diamond29%20%
Design Arena Elo
Speed (tokens/sec)
Latency
Input price /M$0.050$0.027
Output price /M$0.100$0.201
Context window131K131K
CapabilitiesJSONVision