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

Open-Jev 2B (Zefan Cai)

Jev rebuild · by Zefan Cai (@Zefan_Cai)

JevBench v1.4.2.2 score

10.0

Rank #78 of 91 ranked systems.

53.3 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
42.3
calibration
55.3
speed
73.5
cost
28.1

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: Open-Jev 2B — Jev rebuild · Score 10.0 (#78)
  • 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, Open-Jev 2B vs Jev 1.13.0. Intelligence: 42.3 vs 53.1; Calibration: 55.3 vs 76.3; Speed: 73.5 vs 83.3; Cost: 28.1 vs 52.0.50100Intelligence42.3 · 53.1Calibration55.3 · 76.3Speed73.5 · 83.3Cost28.1 · 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, Open-Jev 2B vs Jev 1.13.0. Easy: 100% vs 100%; Standard: 79% vs 99%; Judge: 88% vs 95%; Hard: 43% vs 74%; Sealed: 26% vs 37%.50100Easy100% · 100%Standard79% · 99%Judge88% · 95%Hard43% · 74%Sealed26% · 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), Open-Jev 2B vs Jev 1.13.0. Ambiguous / abstain: 22% vs 43%; Judge: 50% vs 54%; Long policy: 14% vs 44%; Multi-hop: 37% vs 64%; Probability: 29% vs 63%; Temporal / numeric: 21% vs 28%; Trade-off: 34% vs 55%; Routing: 80% vs 100%; Trap / adversarial: 63% vs 83%; Paraphrase: 43% vs 64%; Safety judge: 31% vs 38%.50100Ambiguous /abstain22% · 43%Judge50% · 54%Long policy14% · 44%Multi-hop37% · 64%Probability29% · 63%Temporal /numeric21% · 28%Trade-off34% · 55%Routing80% · 100%Trap /adversarial63% · 83%Paraphrase43% · 64%Safety judge31% · 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, Open-Jev 2B vs Jev 1.13.0. Ambiguous / abstain: 22% vs 30%; Judge: 41% vs 34%; Long policy: 8% vs 28%; Multi-hop: 24% vs 45%; Paraphrase: 43% vs 64%; Probability: 25% vs 50%; Safety judge: 31% vs 38%; Temporal / numeric: 23% vs 29%; Trade-off: 31% vs 38%; Trap / adversarial: 42% vs 42%.50100Ambiguous /abstain22% · 30%Judge41% · 34%Long policy8% · 28%Multi-hop24% · 45%Paraphrase43% · 64%Probability25% · 50%Safety judge31% · 38%Temporal /numeric23% · 29%Trade-off31% · 38%Trap /adversarial42% · 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
MIT (loader); Apache-2.0 (adapter and pinned Qwen base); CC0-1.0 public training projection
Cost evidence
estimated; the board’s row disclosure contains the published basis.
Endpoint condition
our RunPod GPU (H100 80GB HBM3), reached over the internet
Note on this row
The author's pinned LoRA adapter, trained scalar decision head and calibration temperature, served by the author's Open-Jev server with prefix caching off, batch size 1 and 4,096-token limit. Serial requests were measured from Sandy over an SSH tunnel to the H100. Self-host latency receives the standing x2 + 0.15 s adjustment. Cost uses the exact Qwen3.5-9B hosted input tariff for 9B and the same conservative same-family proxy for the unlisted 2B; neither receives an automatic 100. Exact normalized comparison found no JevBench public task state or instruction in the 79,116-row public training projection.
Published source
https://github.com/Zefan-Cai/Open-Jev

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.