modelgrep

Google: Gemini 2.5 Flash Lite vs Nous: Hermes 3 405B Instruct

Google: Gemini 2.5 Flash Lite wins on more metrics (5 of 7), but the right pick depends on what you optimize for — see the breakdown below.

MetricGoogle: Gemini 2.5 Flash LiteNous: Hermes 3 405B Instruct
Intelligence Index17.617.6
Coding Index9.518.1
GPQA Diamond63%54%
Design Arena Elo
Speed (tokens/sec)12122
Latency356ms345ms
Input price /M$0.100$1.00
Output price /M$0.400$1.00
Context window1.0M131K
CapabilitiesReasoningToolsJSONVisionAudioJSON