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

kev 4B (research preview)

Jev rebuild · by Jared Palmer

JevBench v1.4.2.2 score

36.1

Rank #30 of 91 ranked systems.

27.2 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
42.1
calibration
39.6
speed
75.7
cost
61.8

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: kev 4B — Jev rebuild · Score 36.1 (#30)
  • 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, kev 4B vs Jev 1.13.0. Intelligence: 42.1 vs 53.1; Calibration: 39.6 vs 76.3; Speed: 75.7 vs 83.3; Cost: 61.8 vs 52.0.50100Intelligence42.1 · 53.1Calibration39.6 · 76.3Speed75.7 · 83.3Cost61.8 · 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, kev 4B vs Jev 1.13.0. Easy: 100% vs 100%; Standard: 92% vs 99%; Judge: 86% vs 95%; Hard: 42% vs 74%; Sealed: 22% vs 37%.50100Easy100% · 100%Standard92% · 99%Judge86% · 95%Hard42% · 74%Sealed22% · 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), kev 4B vs Jev 1.13.0. Ambiguous / abstain: 12% vs 43%; Judge: 42% vs 54%; Long policy: 22% vs 44%; Multi-hop: 29% vs 64%; Probability: 35% vs 63%; Temporal / numeric: 13% vs 28%; Trade-off: 26% vs 55%; Routing: 100% vs 100%; Trap / adversarial: 73% vs 83%; Paraphrase: 21% vs 64%; Safety judge: 44% vs 38%.50100Ambiguous /abstain12% · 43%Judge42% · 54%Long policy22% · 44%Multi-hop29% · 64%Probability35% · 63%Temporal /numeric13% · 28%Trade-off26% · 55%Routing100% · 100%Trap /adversarial73% · 83%Paraphrase21% · 64%Safety judge44% · 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, kev 4B vs Jev 1.13.0. Ambiguous / abstain: 11% vs 30%; Judge: 32% vs 34%; Long policy: 20% vs 28%; Multi-hop: 13% vs 45%; Paraphrase: 21% vs 64%; Probability: 29% vs 50%; Safety judge: 44% vs 38%; Temporal / numeric: 14% vs 29%; Trade-off: 27% vs 38%; Trap / adversarial: 50% vs 42%.50100Ambiguous /abstain11% · 30%Judge32% · 34%Long policy20% · 28%Multi-hop13% · 45%Paraphrase21% · 64%Probability29% · 50%Safety judge44% · 38%Temporal /numeric14% · 29%Trade-off27% · 38%Trap /adversarial50% · 42%
Share correct within each sealed family — system-level aggregates; the items stay private.

Availability and evidence

Openness
Unknown in the published row
License note
Apache-2.0
Cost evidence
estimated; the board’s row disclosure contains the published basis.
Endpoint condition
our RunPod GPU (GeForce RTX 3090 24 GB, community cloud CA), reached over the internet
Note on this row
Self-hosted from the author's repository at commit 20fa626 through its native TypeSafe-compatible `/v1/systemone` server, BF16 on an RTX 3090; measured serially from Sandy over the internet. The author labels this checkpoint a research preview. 306/308 sealed items answered validly (failures count as wrong)
Published source
https://github.com/jaredpalmer/kev

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.