$ onpar receipt receipt.json


Migration receipt · gpt-4o-mini → qwen3-30b-a3b
issued 2026-11-03 · eval set da030c9c5fb9 · digest 9c476cc9bf93

NOT PROVEN either way — not enough evidence:
  · tools offered, unused · free text · <=1k context 100% [76%–100%] over 12 items — needs 35 to clear the bar at this rate
  · free text · <=1k context #1  92% [80%–97%] over 40 items — needs 525 to clear the bar at this rate
  · tool-calling (2 tools) · empty response · <=1k context 100% [76%–100%] over 12 items (12 duplicate prompt(s) merged) — needs 35 to clear the bar at this rate
  · free text · <=1k context #2  86% [72%–94%] over 37 items (3 duplicate prompt(s) merged)
  · free text · <=1k context #3  100% [88%–100%] over 29 items (11 duplicate prompt(s) merged) — needs 35 to clear the bar at this rate
  · free text · <=1k context #4  100% [21%–100%] over 1 items — needs 35 to clear the bar at this rate

Proven at or above the 90% bar:
  · structured JSON · <=1k context 100% [91%–100%] over 40 items

Movable: 20% of captured traffic
Saving: unknown — no cost rate configured; pass --incumbent-cost and --candidate-cost
Coverage: incomplete — 6 cluster(s) unresolved
Judge: none used — every claim is from deterministic graders
Agentic: same move on 90%–100% of 36 tool-context steps · κ=1.00 · 0 answered where a tool was called, 0 called a tool where an answer was
Drawn from 197 captured requests



# recorded 2026-11-03 on M4 Max · 128 GB · onpar 1.5.2
