... . --> pavilion-weeyuga-v3 — qwen2.5/qwen2.5-coder/qwen3/qwen3.5 on pavilion — benchmarks.weeyuga.com
← All benchmarks
Headline Methodology Results Cold vs warm Raw

Archive · 29 APR 2026 · Pavilion · HP laptop · GTX 1050 4 GB · 16 GB RAM · i7-9750H · qwen2.5/qwen2.5-coder/qwen3/qwen3.5 · chat

pavilion-weeyuga-v3 — qwen2.5/qwen2.5-coder/qwen3/qwen3.5 on pavilion

96 calls across 16 cell(s); 12 errors

Archive run

This run is published for transparency. The site-grade is archive-only — the run was meta-only, didn’t complete cleanly, lacked documented methodology, or had a methodology issue a newer run supersedes. Headline numbers in this row should not be cited as current findings without reading any caveat below and the underlying run.md in the public archive.

Methodology

See SITE_DATA_AUDIT_AND_MIGRATION_PLAN_2026-05-06 for the full procedure. Reproducible at git SHA 371ce70c.

Results

Cell tok/s mean tok/s p50 tok/s p95 duration p50 calls
qwen3.5:4b6
qwen3.5:35b-a3b-uncensored-iq1m6
qwen3.5:35b-a3b-iq2s6
qwen3.5:9b-q6k6
qwen3.5:9b-q4km6
qwen3.5:2b6
qwen3.5:0.8b6
qwen3.5:9b6
qwen2.5-coder:14b6
qwen2.5-coder:3b6
qwen3:14b6
qwen3:8b6
qwen3:4b6
qwen2.5:3b6
qwen2.5-coder:1.5b6
qwen2.5-coder:0.5b6

tokens per second — mean · p50 · p95

No tokens-per-second data captured.

Cold start vs warm

Cold-start measurements are the first call into a model after it loads from disk; warm calls are everything after. The ratio shows how much of the deployment’s wall-time cost is one-time vs steady-state.

Cellcold ncold tok/scold p50warm nwarm tok/swarm p50warm/cold
pavilion:weeyuga:qwen3.5:4b24
pavilion:weeyuga:qwen3.5:35b-a3b-…24
pavilion:weeyuga:qwen3.5:35b-a3b-…24
pavilion:weeyuga:qwen3.5:9b-q6k24
pavilion:weeyuga:qwen3.5:9b-q4km24
pavilion:weeyuga:qwen3.5:2b24
pavilion:weeyuga:qwen3.5:0.8b24
pavilion:weeyuga:qwen3.5:9b24
pavilion:weeyuga:qwen2.5-coder:14b24
pavilion:weeyuga:qwen2.5-coder:3b24
pavilion:weeyuga:qwen3:14b24
pavilion:weeyuga:qwen3:8b24
pavilion:weeyuga:qwen3:4b24
pavilion:weeyuga:qwen2.5:3b24
pavilion:weeyuga:qwen2.5-coder:1.5b24
pavilion:weeyuga:qwen2.5-coder:0.5b24

Raw data

Every run gets its JSONL, log, summary, and metadata published. Clone the archive; re-run it; tell us where we got it wrong.

Cite

Margetic, S. et al. (2026). benchmarks.weeyuga.com/benchmarks/ad057f5b.html
Public benchmarks of the Weeyuga cluster. Run id: ad057f5b-ed3f-4a95-a38e-361be310ffd6. SHA 371ce70c.

Related runs