The LLM benchmark aggregator.

Djev Spark

Other

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Decisions48.2−25.1—/85✱0/1

Routing & classification47.6−26.4—/85✱0/1

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Djev Spark, left for the other.

Mirrorwon 2 · lost 0

Raw Qwen3 0.6B direct logitswon 2 · lost 0

GLiNER2won 2 · lost 0

GLiNER2 largewon 2 · lost 0

Raw Qwen3 1.7B direct logitswon 2 · lost 0

CLM 8Bwon 2 · lost 0

SimpleJevwon 2 · lost 0

GLiNER2.5 smallwon 2 · lost 0

Laya 322Mwon 2 · lost 0

openJev Verdictwon 2 · lost 0

GLiNER2.5 multiwon 2 · lost 0

kev 0.5Bwon 2 · lost 0

Verdict Smallwon 2 · lost 0

JevActwon 2 · lost 0

Smalljev semantic-v9won 2 · lost 0

kev 0.6Bwon 2 · lost 0

Alibaba GTE Reranker ModernBERT-basewon 2 · lost 0

OpenDecisionwon 2 · lost 0

Certo v1won 2 · lost 0

Mixedbread mxbai-rerank-base-v2won 2 · lost 0

BAAI bge-reranker-v2-m3won 2 · lost 0

Open Jev Deberta V3 Largewon 2 · lost 0

Qwen3.5-0.8B Decision Modelwon 2 · lost 0

Typecastlmwon 2 · lost 0

Raw Qwen3 8B direct logitswon 1 · lost 1

Open Jev JSON Canvaswon 1 · lost 1

Raw Qwen3 4B Instruct direct logitswon 1 · lost 1

System Onewon 1 · lost 1

Decider 2Bwon 1 · lost 1

kev 4Bwon 1 · lost 1

kev 8Bwon 1 · lost 1

Spark S1 4B V6won 1 · lost 1

Malkuth 2Bwon 1 · lost 1

Open Alternative Jevwon 1 · lost 1

Layawon 1 · lost 1

openJev Verdict 1.4won 1 · lost 1

Decision Fastwon 1 · lost 1

Jeffwon 1 · lost 1

Lev 350Mwon 1 · lost 1

Vonwon 1 · lost 1

Open-Jev 2Bwon 0 · lost 2

System One Openwon 0 · lost 2

Raw Phi-4 mini direct logitswon 0 · lost 2

OpenSourceJevwon 0 · lost 2

Bespoke Nimble 9Bwon 0 · lost 2

Malkuth 4Bwon 0 · lost 2

SimpleJev Qwen3.6-35B-A3Bwon 0 · lost 2

Open-Jev 9Bwon 0 · lost 2

Qwen3.5-9B Jev-like data-mix v2won 0 · lost 2

Decision Machine 1won 0 · lost 2

Jev Localwon 0 · lost 2

Qwen3 Reranker 4Bwon 0 · lost 2

Jobe Qwen3.5-4Bwon 0 · lost 2

Jev Omniwon 0 · lost 2

SemIfwon 0 · lost 2

Metask Jev 4Bwon 0 · lost 2

Standard One 8Bwon 0 · lost 2

Decision 2Bwon 0 · lost 2

Decider 35B A3Bwon 0 · lost 2

Winnow-12B Q8won 0 · lost 2

JEV Qwen3.5-9B Base NVFP4won 0 · lost 2

Gemini 3.1 Flash Litewon 0 · lost 2

OpenJevwon 0 · lost 2

ZeroEntropy zerank-2won 0 · lost 2

local-jev Qwen3.5-4Bwon 0 · lost 2

Jqvwon 0 · lost 2

reflex 4Bwon 0 · lost 2

Openjev Sglangwon 0 · lost 2

LitJevwon 0 · lost 2

Reflex 27Bwon 0 · lost 2

JevK5 v0.2.0won 0 · lost 2

swanOnewon 0 · lost 2

Decider-4b v2won 0 · lost 2

Cygnetwon 0 · lost 2

JevOnewon 0 · lost 2

SimpleJev Qwen3.8-27Bwon 0 · lost 2

Hopperwon 0 · lost 2

NInfer Qwen3.8-27B NVFP4won 0 · lost 2

NInfer Qwen3.8-Flash-Next mixedwon 0 · lost 2

Jev 1.13.0won 0 · lost 2

Instinctwon 0 · lost 2

Gemma 4 31Bwon 0 · lost 2

Gemma 4 31B IT (Autoloops)won 0 · lost 2

Djevwon 0 · lost 2

GPT-5.6 Lunawon 0 · lost 2

GPT-6 Lunawon 0 · lost 2

DeepSeek V4.1 Flashwon 0 · lost 2

§ 3 · Sources

Where the numbers come from

1 publication, 1 figures. Every one links to the page it was read from.

✱ LargitData — Jev on multi-turn RAG routing ↗1 measureread 2026-09-26

Decision correct 32.2%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=djev-spark)](https://publicai.io/model-index/m/djev-spark)