Contributors › DandinPower's local-LLM benchmark configs

DandinPower's local-LLM benchmark configs

Every tracked benchmark config measured by DandinPower: model, quant, backend, tokens per second, source-linked and trust-tiered.

Snapshot 2026-09-30 · 254 configs · 116 models. Every number is generated from the tracker snapshot and linked to where it was measured; nothing here is typed by hand.

7 configs · 7 models · 2 stars · 146.1 tok/s best single-stream · DandinPower on github.

Configs measured by DandinPower

Model	Quant	Backend	Decode	Context	Trust	Source
Qwen3 1.7B	Q4_K_M	llama.cpp	146.1 tok/s	2048	Extracted, source-linked	source
Ministral 3B	Q4_K_M	llama.cpp	86.6 tok/s	2048	Extracted, source-linked	source
Qwen3 30B (MoE)	Q4_K_M	llama.cpp	83.8 tok/s	2048	Extracted, source-linked	source
Qwen3 8B	Q4_K_M	llama.cpp	42 tok/s	2048	Extracted, source-linked	source
Ministral 8B	Q4_K_M	llama.cpp	40.1 tok/s	2048	Extracted, source-linked	source
Ministral 14B	Q4_K_M	llama.cpp	25.7 tok/s	2048	Extracted, source-linked	source
Qwen3 32B (Dense)	Q4_K_M	llama.cpp	10.5 tok/s	2048	Extracted, source-linked	source

More: all configs · hardware · methodology · llms.txt · llms-full.txt · API (OpenAPI) · JSON snapshot. Built by Altronis, private on-prem AI, Singapore.