DATASET VIEW
The numbers
without the gloss.
A living snapshot of 175 community reports. Every number is grounded in what contributors actually shared — including the gaps.
Power is a budget. Quality is a practice.
175community reports
148machines described
155GPUs identified
4832 GBknown total VRAM
What the dataset can tell us
coverage, not a benchmark77%of reports include identifiable hardware
26%report a generation speed
3community snapshots beyond the genesis set
Cloud replacement
n = 175By reported capacity
derived from exact GB where known8 GB or less
8
9–12 GB
6
13–16 GB
10
17–20 GB
2
21–24 GB
14
25–32 GB
13
33–48 GB
3
49–96 GB
9
97 GB+
11
Reported detail
share of reports that state each fieldMost mentioned models
n = mentionsQwen 3.8 27B60
Qwen 3.8 Flash Next27
Qwen 3.6 35B-A3B19
DeepSeek V4 Flash10
Qwen5
Gemma 4 12B4
Runtime signals
n = reportsllama.cpp15
vLLM6
Ollama5
LM Studio5
nInfer4
Strata4
Tools / workflow
n = reportsHermes12
OpenCode7
Pi3
Cline in VS Code2
Cline1
Orca1
Snapshot origins
n = entriesMulti-machine reports: 9. How badges and XP work.
Use cases represented
n = entries · click to exploreTotals can grow as the community adds detail. Missing hardware is preserved as missing — never inferred.