Benchmarks
More
OverviewBenchmaxxingCompareChartsCost vs CapabilityProviders per ModelProvider explorerGatewaysEU & SovereignJev-class modelsAbout
JevBench by Benchmark Heaven · v1.4.2 · individual system

verdict-small (Manavarya09, multilingual-e5-small 118M)

This detail uses the public, hash-checked v1.4.2 aggregate. Scores and ranks can change when a new release is published; the page preview remains name-only.

JevBench v1.4.2 score

5.7

Rank #82 of 89 ranked systems.

58.4 points behind Jev 1.13.0's 64.1.

Where it sits among the 89 ranked systems. The marked tick is Jev 1.13.0 (63.3).

Published axes

intelligence
18.1
calibration
66.9
speed
85.0
cost
96.6

Accuracy per tier, incl. sealed

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: verdict-small — Jev rebuild · Score 5.7 (#82)
  • B: decider-4b v2 — system-one-open · Score 64.1 (#1)

The four score axes

Radar: the four score axes, two systemsThe four score axes, verdict-small vs decider-4b v2. Intelligence: 18.1 vs 49.4; Calibration: 66.9 vs 75.0; Speed: 85.0 vs 92.9; Cost: 96.6 vs 60.9.50100Intelligence18.1 · 49.4Calibration66.9 · 75.0Speed85.0 · 92.9Cost96.6 · 60.9
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, verdict-small vs decider-4b v2. Easy: 89% vs 100%; Standard: 51% vs 97%; Judge: 28% vs 88%; Hard: 40% vs 67%; Sealed: 29% vs 35%.50100Easy89% · 100%Standard51% · 97%Judge28% · 88%Hard40% · 67%Sealed29% · 35%
Share correct per tier; Sealed = the 308 private decisions, aggregate only.

Hard tier by family (v1.2 topics)

decider-4b v2 has no published v1.2 hard-tier family breakdown.

Radar: hard tier by family (v1.2 topics), two systemsHard tier by family (v1.2 topics), verdict-small vs decider-4b v2. Adversarial: 42% vs —; Ambiguous: 29% vs —; Judge: 45% vs —; Long policy: 45% vs —; Multi-hop: 26% vs —; Probability: 25% vs —; Routing: 80% vs —; Temporal / numeric: 30% vs —; Trade-off: 67% vs —; Trap: 50% vs —.50100Adversarial42%Ambiguous29%Judge45%Long policy45%Multi-hop26%Probability25%Routing80%Temporal /numeric30%Trade-off67%Trap50%
Share correct within each family of the 47 v1.2 hard-tier decisions (public and held-out).

Sealed set by family

Radar: sealed set by family, two systemsSealed set by family, verdict-small vs decider-4b v2. Ambiguous / abstain: 19% vs 43%; Judge: 68% vs 32%; Long policy: 20% vs 30%; Multi-hop: 16% vs 34%; Paraphrase: 50% vs 50%; Probability: 36% vs 36%; Safety judge: 50% vs 38%; Temporal / numeric: 14% vs 30%; Trade-off: 15% vs 35%; Trap / adversarial: 25% vs 33%.50100Ambiguous /abstain19% · 43%Judge68% · 32%Long policy20% · 30%Multi-hop16% · 34%Paraphrase50% · 50%Probability36% · 36%Safety judge50% · 38%Temporal /numeric14% · 30%Trade-off15% · 35%Trap /adversarial25% · 33%
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 (verdictml code and the Manav2op/verdict-small checkpoint over intfloat/multilingual-e5-small)
Cost evidence
estimated; the board’s row disclosure contains the published basis.
Endpoint condition
our CPU (8 threads, Ryzen 5 3600)
Published source
https://github.com/Manavarya09/verdict

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