Benchmarks
More
OverviewBenchmaxxingCompareChartsCost vs CapabilityProviders per ModelProvider explorerGatewaysEU & SovereignJev-class modelsImage JevBenchAbout
JevBench by Benchmark Heaven · v1.4.2.2 · individual system

Mirror

Jev rebuild · by Bluusun

JevBench v1.4.2.2 score

2.1

Rank #86 of 91 ranked systems.

61.2 points behind Jev 1.13.0's 63.3.

Where it sits among the 91 ranked systems. The marked tick is Jev 1.13.0 (63.3).

Published axes

intelligence
13.6
calibration
26.0
speed
70.8
cost
73.3

Against Jev 1.13.0

Four radars compare this fixed pair across the score axes, accuracy per tier, and accuracy by family on the hard tier and sealed set. Further out is better on every spoke.

  • A: Mirror — Jev rebuild · Score 2.1 (#86)
  • B: Jev 1.13.0 — Jev (TypeSafe, closed) · Score 63.3 (#4)

The four score axes

Radar: the four score axes, two systemsThe four score axes, Mirror vs Jev 1.13.0. Intelligence: 13.6 vs 53.1; Calibration: 26.0 vs 76.3; Speed: 70.8 vs 83.3; Cost: 73.3 vs 52.0.50100Intelligence13.6 · 53.1Calibration26.0 · 76.3Speed70.8 · 83.3Cost73.3 · 52.0
0–100, the values in the table. A label-only system has no calibration (counted as 0).

Accuracy per tier, incl. sealed

Radar: accuracy per tier, incl. sealed, two systemsAccuracy per tier, incl. sealed, Mirror vs Jev 1.13.0. Easy: 88% vs 100%; Standard: 46% vs 99%; Judge: 32% vs 95%; Hard: 17% vs 74%; Sealed: 9% vs 37%.50100Easy88% · 100%Standard46% · 99%Judge32% · 95%Hard17% · 74%Sealed9% · 37%
Share correct per tier; Sealed = the 308 private decisions, aggregate only.

Current question set by family (hard + sealed)

Radar: current question set by family (hard + sealed), two systemsCurrent question set by family (hard + sealed), Mirror vs Jev 1.13.0. Ambiguous / abstain: 12% vs 43%; Judge: 39% vs 54%; Long policy: 0% vs 44%; Multi-hop: 3% vs 64%; Probability: 13% vs 63%; Temporal / numeric: 9% vs 28%; Trade-off: 3% vs 55%; Routing: 10% vs 100%; Trap / adversarial: 30% vs 83%; Paraphrase: 7% vs 64%; Safety judge: 0% vs 38%.50100Ambiguous /abstain12% · 43%Judge39% · 54%Long policy0% · 44%Multi-hop3% · 64%Probability13% · 63%Temporal /numeric9% · 28%Trade-off3% · 55%Routing10% · 100%Trap /adversarial30% · 83%Paraphrase7% · 64%Safety judge0% · 38%
Share correct per family across the 220 hard-tier decisions (public and held out) and the 308 sealed decisions of v1.4, pooled; Routing is hard-tier only, Paraphrase and Safety judge sealed only.

Sealed set by family

Radar: sealed set by family, two systemsSealed set by family, Mirror vs Jev 1.13.0. Ambiguous / abstain: 11% vs 30%; Judge: 27% vs 34%; Long policy: 0% vs 28%; Multi-hop: 3% vs 45%; Paraphrase: 7% vs 64%; Probability: 14% vs 50%; Safety judge: 0% vs 38%; Temporal / numeric: 9% vs 29%; Trade-off: 0% vs 38%; Trap / adversarial: 17% vs 42%.50100Ambiguous /abstain11% · 30%Judge27% · 34%Long policy0% · 28%Multi-hop3% · 45%Paraphrase7% · 64%Probability14% · 50%Safety judge0% · 38%Temporal /numeric9% · 29%Trade-off0% · 38%Trap /adversarial17% · 42%
Share correct within each sealed family — system-level aggregates; the items stay private.

Availability and evidence

Openness
Code and weights marked open in the published row
License note
Apache-2.0 wrapper and Mirror release; upstream DeBERTa/model assets retain their own terms
Cost evidence
estimated; the board’s row disclosure contains the published basis.
Endpoint condition
hosted submission endpoint, serial from Germany; strict 512-token context rejection
Note on this row
171 intended HTTP 422 context rejections (over 512-token limit) counted once each as misses; only 363/534 valid distributions returned. Found via Gmail submission (Lewis). 102/308 sealed items answered validly (failures count as wrong)
Published source
https://github.com/Bluusun/Decision-API

From the public v1.4.2.2 aggregate. Scores and ranks can change when a new release is published.

Read the full board and published method. The overall score is a composite, not raw accuracy.