AI Mode Benchmarks

AI MODE
TEST CASES

Each model gets a full case study: ON/OFF proof, local screenshots, tokens per second and the exact kind of workload GenesiOS is tuning for.

CHAT - LOCAL LLM - CPU TEST01

Mistral 7b

General local chat test with prompt generation, sustained tokens and the full AI Mode profile applied while inference is running.

9 -> 41 tokens/s+355% upliftperformance governor
OFF
9
tokens/s
ON
41
tokens/s
GAIN
+355%
genesi-ai-mode
+355%
Mistral 7b AI Mode ON placeholder
Mistral 7b AI Mode OFF placeholder
AI Mode OFF
AI Mode ON
CODE - AGENT TERMINAL - GENESI CODE02

Qwen Coder

Code-oriented completion test for Genesi Code style workloads, where scheduling priority and memory tuning matter more.

5.2 -> 8.7 tokens/s+29% uplifthuge pages enabled
OFF
5.2
tokens/s
ON
8.7
tokens/s
GAIN
+67%
genesi-ai-mode
+67%
Qwen Coder AI Mode ON placeholder
Qwen Coder AI Mode OFF placeholder
AI Mode OFF
AI Mode ON