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

spark-s1-4b-v6 (Open Spark Jev, abhishek085)

Jev rebuild · by Abhishek Rai (abhishek085)

JevBench v1.4.2.2 score

44.6

Rank #17 of 91 ranked systems.

18.7 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
45.1
calibration
47.6
speed
81.0
cost
57.9

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.

The four score axes

Radar: the four score axes, two systemsThe four score axes, spark-s1-4b-v6 vs Jev 1.13.0. Intelligence: 45.1 vs 53.1; Calibration: 47.6 vs 76.3; Speed: 81.0 vs 83.3; Cost: 57.9 vs 52.0.50100Intelligence45.1 · 53.1Calibration47.6 · 76.3Speed81.0 · 83.3Cost57.9 · 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, spark-s1-4b-v6 vs Jev 1.13.0. Easy: 100% vs 100%; Standard: 98% vs 99%; Judge: 92% vs 95%; Hard: 60% vs 74%; Sealed: 27% vs 37%.50100Easy100% · 100%Standard98% · 99%Judge92% · 95%Hard60% · 74%Sealed27% · 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), spark-s1-4b-v6 vs Jev 1.13.0. Ambiguous / abstain: 25% vs 43%; Judge: 51% vs 54%; Long policy: 28% vs 44%; Multi-hop: 49% vs 64%; Probability: 46% vs 63%; Temporal / numeric: 17% vs 28%; Trade-off: 39% vs 55%; Routing: 100% vs 100%; Trap / adversarial: 75% vs 83%; Paraphrase: 43% vs 64%; Safety judge: 44% vs 38%.50100Ambiguous /abstain25% · 43%Judge51% · 54%Long policy28% · 44%Multi-hop49% · 64%Probability46% · 63%Temporal /numeric17% · 28%Trade-off39% · 55%Routing100% · 100%Trap /adversarial75% · 83%Paraphrase43% · 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, spark-s1-4b-v6 vs Jev 1.13.0. Ambiguous / abstain: 19% vs 30%; Judge: 34% vs 34%; Long policy: 18% vs 28%; Multi-hop: 21% vs 45%; Paraphrase: 43% vs 64%; Probability: 39% vs 50%; Safety judge: 44% vs 38%; Temporal / numeric: 18% vs 29%; Trade-off: 31% vs 38%; Trap / adversarial: 33% vs 42%.50100Ambiguous /abstain19% · 30%Judge34% · 34%Long policy18% · 28%Multi-hop21% · 45%Paraphrase43% · 64%Probability39% · 50%Safety judge44% · 38%Temporal /numeric18% · 29%Trade-off31% · 38%Trap /adversarial33% · 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 (code and weights); base Qwen/Qwen3.5-4B Apache-2.0
Cost evidence
estimated; the board’s row disclosure contains the published basis.
Endpoint condition
our RunPod GPU (L40 48 GB, Czechia), reached over the internet from Germany
Note on this row
Abhishek085/spark-s1-4b-v6 revision 93d49ddbfb29212e3296635a75a3e80cf69da027, code github.com/abhishek085/open-spark-jev 30ac6d89b7fa36c644cf86aac68f35c1d276a919, the author's own MenuScorer.decide with his fitted calibration.json temperature, bf16, base Qwen/Qwen3.5-4B, transformers 5.17.0 / torch 2.8.0 from the pod image, flash-linear-attention 0.5.2 installed, causal_conv1d not installable here (no wheel builds against this toolchain), on our RunPod L40 in Czechia Offline local open-weight inference on the frozen 308-item v1.4 set in a network-disabled, read-only container; no operator endpoint received sealed text.
Published source
https://github.com/abhishek085/open-spark-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.