75 models · updated 2 November 2026

Decision models from 500M to 2B parameters

The middle tier: more capacity than an encoder, still cheap enough to self-host without a serious GPU.

Also in this dimensionUnder 500M 126Over 2B 114
Community card

decider-2b

Mapika

The most-downloaded open-weight decision model in the post-Jev wave. A Qwen3.5-2B decoder fine-tuned to emit typed decisions, and the largest member of a three-model family.

1.9B · Qwen/Qwen3.5-2B-Base · apache-2.0

Community card

agent-jev

aimeigaoshou

Qwen3-0.6B fine-tuned to return typed decisions, 598M parameters.

598M · Qwen/Qwen3-0.6B · apache-2.0

Community card

OpenThai-SystemOne

iapp

Qwen3.5-0.8B-Base fine-tuned to return typed decisions, 753M parameters.

753M · Qwen/Qwen3.5-0.8B-Base · apache-2.0

Community card

Jev-Style-Qwen3.5-2B-Decision-GGUF

chaoliangUNSW

A GGUF build of Jev-Style-Qwen3.5-2B-Decision, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · Qwen/Qwen3.5-2B-Base · apache-2.0

Community card

Jev-Style-0.8B-Decision-v3-GGUF

chaoliangUNSW

A GGUF build of Jev-Style-0.8B-Decision-v3, for running typed decisions on llama.cpp instead of a hosted API.

752M · chaoliangUNSW/Jev-Style-0.8B-Decision-v3 · apache-2.0

Community card

jev

feder-cr

jev, an open-weight model published to return typed decisions, 882M parameters.

882M · mit

Community card

Jev-Style-2B-Decision-v3

chaoliangUNSW

Qwen3.5-2B fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Qwen/Qwen3.5-2B · apache-2.0

Community card

Jev-Style-2B-Decision-v3-GGUF

chaoliangUNSW

A GGUF build of Jev-Style-2B-Decision-v3, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · chaoliangUNSW/Jev-Style-2B-Decision-v3 · apache-2.0

Community card

decider-0.8b

Mapika

Qwen3.5-0.8B-Base fine-tuned to return typed decisions, 752M parameters.

752M · Qwen/Qwen3.5-0.8B-Base · apache-2.0

Community card

Jev-Style-Qwen3.5-2B-Decision-v2-GGUF

chaoliangUNSW

A GGUF build of Jev-Style-Qwen3.5-2B-Decision-v2, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-v2 · apache-2.0

Community card

JevK5-GGUF

alibiserikbay

A GGUF build of JevK5, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · alibiserikbay/JevK5 · apache-2.0

Community card

Jev-Style-Qwen3.5-2B-Decision-MLX-bf16

chaoliangUNSW

An MLX build of Jev-Style-Qwen3.5-2B-Decision, for running typed decisions on Apple Silicon instead of a hosted API.

1.9B · Qwen/Qwen3.5-2B-Base · apache-2.0

Community card

Tiny-Jev

lostargon

Qwen3-0.6B fine-tuned to return typed decisions, 596M parameters.

596M · Qwen/Qwen3-0.6B · apache-2.0

Community card

qwen3-0.6b-rlcd-decision

anthonym21

Qwen3-0.6B-Base fine-tuned to return typed decisions, 596M parameters.

596M · Qwen/Qwen3-0.6B-Base · apache-2.0

Community card

jev-novel-2.5-0.8b-bf16

aikexue170

Qwen3.5-0.8B fine-tuned to return typed decisions, 752M parameters.

752M · Qwen/Qwen3.5-0.8B · apache-2.0

Community card

decider-2b-GGUF

Mapika

A GGUF build of decider-2b, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · Mapika/decider-2b · apache-2.0

Community card

decider-2b-vision-GGUF

mindchain

A GGUF build of decider-2b-vision, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · Mapika/decider-2b-vision · apache-2.0

Community card

JevEmbed-Qwen3-Embedding-0.6B

HIT-TMG

Qwen3-Embedding-0.6B fine-tuned to return typed decisions, 596M parameters.

596M · Qwen/Qwen3-Embedding-0.6B · apache-2.0

Community card

jev-control-core

abhishek085

A GGUF build of Qwen3.5-0.8B-Base, for running typed decisions on llama.cpp instead of a hosted API.

752M · Qwen/Qwen3.5-0.8B-Base · apache-2.0

Community card

decider-2b-dwq-4bit

midium-ai

decider-2b fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Mapika/decider-2b · apache-2.0

Community card

laya-v

henilchopada

laya fine-tuned to return typed decisions, 752M parameters.

752M · convaiinnovations/laya · apache-2.0

Community card

depguard-jev-style-1.5b

JackKozmo29

Qwen2.5-1.5B-Instruct fine-tuned to return typed decisions, 1.5B parameters.

1.5B · Qwen/Qwen2.5-1.5B-Instruct · apache-2.0

Community card

mailguard-jev-style-1.5b

JackKozmo29

Qwen2.5-1.5B-Instruct fine-tuned to return typed decisions, 1.5B parameters.

1.5B · Qwen/Qwen2.5-1.5B-Instruct · apache-2.0

Community card

sevclassifier-jev-style-1.5b

JackKozmo29

Qwen2.5-1.5B-Instruct fine-tuned to return typed decisions, 1.5B parameters.

1.5B · Qwen/Qwen2.5-1.5B-Instruct · apache-2.0

Community card

codeguard-jev-style-1.5b

JackKozmo29

Qwen2.5-1.5B-Instruct fine-tuned to return typed decisions, 1.5B parameters.

1.5B · Qwen/Qwen2.5-1.5B-Instruct · apache-2.0

Community card

EXAONE-4.0-1.2B-JEV-GGUF

mradermacher

A GGUF build of EXAONE-4.0-1.2B-JEV, for running typed decisions on llama.cpp instead of a hosted API.

1.3B · carrtesy/EXAONE-4.0-1.2B-JEV · other

Community card

JevK5-2B

alibiserikbay

Qwen3.5-2B fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Qwen/Qwen3.5-2B · apache-2.0

Community card

OpenThai-SystemOne-GGUF

iapp

A GGUF build of OpenThai-SystemOne, for running typed decisions on llama.cpp instead of a hosted API.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-Ollama

iapp

OpenThai-SystemOne fine-tuned to return typed decisions, 752M parameters.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

Jev-Style-2B-Decision-v3-GGUF

mradermacher

A GGUF build of Jev-Style-2B-Decision-v3, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · chaoliangUNSW/Jev-Style-2B-Decision-v3 · apache-2.0

Community card

jev-decisions-v1-model

GhostScientist

Qwen3.5-0.8B fine-tuned to return typed decisions, 752M parameters.

752M · Qwen/Qwen3.5-0.8B · apache-2.0

Community card

Erabo-2B

tanakaisworking

decider-2b fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Mapika/decider-2b · apache-2.0

Community card

OOMU-SystemOne-0.6B-GGUF

oomu

OOMU-SystemOne-0.6B-GGUF, an open-weight model published to return typed decisions, 596M parameters.

596M · apache-2.0

Community card

Gemma-1B-Jev-Decision-Classification-SystemOne

mondk

gemma-3-1b-pt fine-tuned to return typed decisions, 1000M parameters.

1000M · google/gemma-3-1b-pt · gemma

Community card

decider-0.8b-fp8

llmtech

A quantised build of decider-0.8b, for running typed decisions on a local runtime instead of a hosted API.

752M · Mapika/decider-0.8b · apache-2.0

Community card

elythros-decider-0.6b

Elythros

Qwen3-0.6B fine-tuned to return typed decisions, 596M parameters.

596M · Qwen/Qwen3-0.6B · apache-2.0

Community card

Jev-LCT-Qwen2.5-1.5B

CaoHaoWei

Jev-LCT-Qwen2.5-1.5B, an open-weight model published to return typed decisions, 1.5B parameters.

1.5B · apache-2.0

Community card

decider-2b-fp8

llmtech

A quantised build of decider-2b, for running typed decisions on a local runtime instead of a hosted API.

1.9B · Mapika/decider-2b · apache-2.0

Community card

archai-jev-qwen-1.5b

nickprock

Qwen2.5-1.5B-Instruct fine-tuned to return typed decisions, 1.5B parameters.

1.5B · Qwen/Qwen2.5-1.5B-Instruct

Community card

edceleste-decider-0.8b-GGUF

mradermacher

A GGUF build of edceleste-decider-0.8b, for running typed decisions on llama.cpp instead of a hosted API.

752M · IqSantiagow/edceleste-decider-0.8b · apache-2.0

Community card

strands-decider-2B-hobson-v19-GGUF

fabricant451

A GGUF build of strands-decider-2B-hobson-v19, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · StrandsAgents/strands-decider-2B-hobson-v19 · apache-2.0

Community card

Jev-Style-Qwen3.5-2B-Decision-v2-MLX-bf16

chaoliangUNSW

An MLX build of Jev-Style-Qwen3.5-2B-Decision-v2, for running typed decisions on Apple Silicon instead of a hosted API.

1.9B · chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-v2 · apache-2.0

Community card

elythros-decider-0.6b-GGUF

mradermacher

A GGUF build of elythros-decider-0.6b, for running typed decisions on llama.cpp instead of a hosted API.

596M · Elythros/elythros-decider-0.6b · apache-2.0

Community card

EXAONE-4.0-1.2B-JEV

carrtesy

EXAONE-4.0-1.2B fine-tuned to return typed decisions, 1.3B parameters.

1.3B · LGAI-EXAONE/EXAONE-4.0-1.2B · other

Community card

archai-jev-games-1.5b-v2

nickprock

archai-jev-qwen-1.5b fine-tuned to return typed decisions, 1.5B parameters.

1.5B · nickprock/archai-jev-qwen-1.5b · apache-2.0

Community card

Jev-Style-Qwen3.5-2B-Decision-v2

chaoliangUNSW

Qwen3.5-2B-Base fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Qwen/Qwen3.5-2B-Base · apache-2.0

Community card

decider-2b-GGUF

DreamBlooms

A GGUF build of decider-2b, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · Mapika/decider-2b · apache-2.0

Community card

decider-0.8b-GGUF

seongukjeong

A GGUF build of decider-0.8b, for running typed decisions on llama.cpp instead of a hosted API.

752M · Mapika/decider-0.8b · apache-2.0

Community card

decider-0.8b-Q8

ma77145

A GGUF build of decider-0.8b, for running typed decisions on llama.cpp instead of a hosted API.

752M · Mapika/decider-0.8b · apache-2.0

Community card

Tiny-Jev-1.7B

lostargon

Qwen3-1.7B fine-tuned to return typed decisions, 1.7B parameters.

1.7B · Qwen/Qwen3-1.7B · apache-2.0

Community card

jev-novel-2.5-0.8b-GGUF

aikexue170

A GGUF build of jev-novel-2.5-0.8b-bf16, for running typed decisions on llama.cpp instead of a hosted API.

752M · aikexue170/jev-novel-2.5-0.8b-bf16 · apache-2.0

Community card

decider-2b-GGUF

Jeetu456

A GGUF build of decider-2b, for running typed decisions on llama.cpp instead of a hosted API.

1.9B · Jeetu456/decider-2b · apache-2.0

Community card

aemeath-systemone-kev

rushsinging

aemeath-systemone-kev, an open-weight model published to return typed decisions, 752M parameters.

752M

Community card

decider-0.8b-GGUF

DreamBlooms

A GGUF build of decider-0.8b, for running typed decisions on llama.cpp instead of a hosted API.

752M · Mapika/decider-0.8b · apache-2.0

Community card

decider-0.8b-GGUF

ardana-ai

A GGUF build of decider-0.8b, for running typed decisions on llama.cpp instead of a hosted API.

752M · Mapika/decider-0.8b · apache-2.0

Community card

decider-0.8b-GGUF

mindchain

A GGUF build of decider-0.8b, for running typed decisions on llama.cpp instead of a hosted API.

752M · Mapika/decider-0.8b · apache-2.0

Community card

decider-2b

nativ-community

decider-2b fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Mapika/decider-2b · apache-2.0

Community card

Qwen3.5-0.8B-Typed-Decisions

SnowGemma

Qwen3.5-0.8B fine-tuned to return typed decisions, 873M parameters.

873M · Qwen/Qwen3.5-0.8B · apache-2.0

Community card

decider-2b

Jeetu456

Qwen3.5-2B-Base fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Qwen/Qwen3.5-2B-Base · apache-2.0

Community card

decider-2b-homelab-router

daviestechlabs

decider-2b fine-tuned to return typed decisions, 1.9B parameters.

1.9B · Mapika/decider-2b · apache-2.0

Community card

OpenThai-SystemOne-MLX-mxfp4

iapp

An MLX build of OpenThai-SystemOne, for running typed decisions on Apple Silicon instead of a hosted API.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-MLX-bf16

iapp

An MLX build of OpenThai-SystemOne, for running typed decisions on Apple Silicon instead of a hosted API.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

plek-1

Drenel

A typed-decision model from Drenel, 783M parameters.

783M · apache-2.0

Community card

OpenThai-SystemOne-bnb-4bit

iapp

OpenThai-SystemOne fine-tuned to return typed decisions, 753M parameters.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-MLX-8bit

iapp

An MLX build of OpenThai-SystemOne, for running typed decisions on Apple Silicon instead of a hosted API.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

JEV-Qwen3.5-0.8B-Base

WIlfLin

Qwen3.5-0.8B fine-tuned to return typed decisions, 873M parameters.

873M · Qwen/Qwen3.5-0.8B · apache-2.0

Community card

OpenThai-SystemOne-MLX-nvfp4

iapp

An MLX build of OpenThai-SystemOne, for running typed decisions on Apple Silicon instead of a hosted API.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-W8A8-INT8

iapp

A quantised build of OpenThai-SystemOne, for running typed decisions on a local runtime instead of a hosted API.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-FP8-Dynamic

iapp

A quantised build of OpenThai-SystemOne, for running typed decisions on a local runtime instead of a hosted API.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-NVFP4A16

iapp

OpenThai-SystemOne fine-tuned to return typed decisions, 753M parameters.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-MLX-4bit

iapp

An MLX build of OpenThai-SystemOne, for running typed decisions on Apple Silicon instead of a hosted API.

752M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-W4A16-GPTQ

iapp

A quantised build of OpenThai-SystemOne, for running typed decisions on a local runtime instead of a hosted API.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-bnb-8bit

iapp

OpenThai-SystemOne fine-tuned to return typed decisions, 753M parameters.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

OpenThai-SystemOne-NVFP4

iapp

OpenThai-SystemOne fine-tuned to return typed decisions, 753M parameters.

753M · iapp/OpenThai-SystemOne · apache-2.0

Community card

LFM2.5-1.2B-RLCD

goutam

LFM2.5-1.2B-Base fine-tuned to return typed decisions, 1.2B parameters.

1.2B · LiquidAI/LFM2.5-1.2B-Base · other

Browse another way

Every model

The full directory, with every facet listed in one place.

What is a decision model?

Start here if the category is new to you.

Find out what your data actually supports

We turn your historical decisions into a frozen test set, then tell you which model, which threshold, and how much of it you can safely automate.

Request a benchmark
Request a benchmark

New benchmark results, methodology notes and model additions. No product marketing.

[email protected]
Models
All modelsGGUF buildsApple Silicon (MLX)Under 500M
Compare
All comparisonsOpen alternatives to JevDecision models vs LLMs
Built with
All projectsDeveloper toolingAI agentsBrowser
Self-host
Compatible serversllama.cppApple SiliconONNX
Browse
Commercial useRestricted licenceQuantised buildsBeyond English
Learn
What is a decision model?Decision models vs LLMsJev benchmark resultsHow to automate a decisionSet a confidence threshold
Company
Request a benchmark

© 2026 DecisionEval · Data updated 2 November 2026

Public benchmark results on this site are produced on frozen, documented test sets, with the method stated on each model page. They describe performance on those specific tasks only and are not a guarantee of performance on your own data. Vendor-reported figures are labelled as such and are not independently verified unless stated.