JEV-9B-MLX-8bit by Nativ: Open-Weight Model

Runs On

What it takes to serve JEV-9B-MLX-8bit (9B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

Precision	Weights	Memory needed	Cheapest setup	Per hour	Also fits
16-bit	17.9 GB	21.5 GB	1x MI300X (192 GB) Vultr	$1.85	1x H100 $1.99 · 1x MI325X $2.00
8-bit	9.0 GB	10.7 GB	1x MI300X (192 GB) Vultr	$1.85	1x H100 $1.99 · 1x MI325X $2.00
4-bit	4.5 GB	5.4 GB	1x MI300X (192 GB) Vultr	$1.85	1x H100 $1.99 · 1x MI325X $2.00

Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 18, 2026.

JEV-9B-MLX-8bit on every accelerator the SAVRN Index prices, at every precision

Model Card

By Nativ, published under apache-2.0, revision c0eafb30e090.

MLX conversion of autotrust/JEV-9B for mlx-vlm, with the System 1 LoRA (including its lmhead update) merged and lmhead, the decision readout, kept unquantized. JEV is a decision model: it returns a probability for every option of a question in one forward pass and does not generate text. JEV support is not yet in an mlx-vlm release

Read Nativ's full model card

MLX conversion of autotrust/JEV-9B for mlx-vlm, with the System 1 LoRA (including its lm_head update) merged and lm_head, the decision readout, kept unquantized. JEV is a decision model: it returns a probability for every option of a question in one forward pass and does not generate text.

JEV support is not yet in an mlx-vlm release:

pip install "git+https://github.com/Lazarus-931/mlx-vlm.git@feat/jev"

from mlx_vlm import load, predict

model, processor = load("nativ-community/JEV-9B-MLX-8bit")
result = predict(model, processor, "The parcel arrived damaged and I want my money back.", {
    "refund": {"type": "bool", "instructions": "Is the customer asking for a refund?"},
})
print(result["answers"]["refund"]["value"])

Field	Value
Source	`autotrust/JEV-9B`
Source revision	`b63f651ce8ed64481d3f5e73ecdb05f740042f01`
Quantization	`affine 8-bit, group size 64`
mlx-vlm	`Lazarus-931/mlx-vlm@feat/jev`
Verification	Same answer as autotrust's System 1 recipe (bf16 backbone + adapter_vllm merged with peft, option-token logits + bias, per-kind temperature) on 4/4 test questions (bool, choice, score, JSON state, 20 options in one pass); identical token ids on 4/4 forward passes; largest probability gap 0.0077

Configuration

Architecture: Qwen3_5ForCausalLM
Context length (tokens): 262,144
Layers: 32
Hidden size: 4,096
Feed-forward size: 12,288
Attention heads: 16
Key/value heads: 4
Head dimension: 256
Vocabulary size: 248,320
Model type: jev_text

Identity and Version

Repository: nativ-community/JEV-9B-MLX-8bit
Publisher: Nativ
Task: Text generation
Modality: Text
Library: mlx
Parameters: 9B parameters
Languages: mlx, mlx-vlm, jev
Revision: c0eafb30e090cb94cff6a762bd2c67baba29f131
First published: 2026-10-08
Last updated: 2026-10-08

Files and Weights

9 files, 10.5 GB in total. The weights are 2 files totalling 10.5 GB in safetensors.

Weights2 files · 10.5 GB

Configuration2 files · 99.5 KB

Tokenizer2 files · 20.0 MB

Documentation1 file · 1.6 KB

Other1 file · 7.8 KB

Repository1 file · 1.6 KB

Every file

File	Type	Size	SHA-256
model-00001-of-00002.safetensors	Weights	5.3 GB	ff9e8ff3f4fc
model-00002-of-00002.safetensors	Weights	5.1 GB	280496e20d64
config.json	Configuration	4.6 KB	—
model.safetensors.index.json	Configuration	94.9 KB	—
README.md	Documentation	1.6 KB	—
chat_template.jinja	Other	7.8 KB	—
.gitattributes	Repository	1.6 KB	—
tokenizer.json	Tokenizer	20.0 MB	06b9509352d2
tokenizer_config.json	Tokenizer	1.1 KB	—

License and Download

License: apache-2.0
Access: Open weights, no gate
Download size: 10.5 GB

Download from Nativ

Released by Nativ through its official repository on Hugging Face. Read the license.

Built From

Derived from autotrust/JEV-9B
Quantized from autotrust/JEV-9B

Memory Requirements

Precision	Weights in memory
As published	10.5 GB
16-bit	17.9 GB
8-bit	9.0 GB
4-bit	4.5 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About JEV-9B-MLX-8bit

How much GPU memory does JEV-9B-MLX-8bit need?

About 21.5 GB at 16-bit and 5.4 GB at 4-bit: the weights (9B parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run JEV-9B-MLX-8bit on?

At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use JEV-9B-MLX-8bit commercially?

Yes. JEV-9B-MLX-8bit is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

What is JEV-9B-MLX-8bit's context length?

262,144 tokens, from the maximum position embeddings in its published configuration.

Similar Models

Model · Text generation

dQwen3.5-9B-Base

IFML

A masked diffusion language model adapted from Qwen3.5-9B. The backbone is hybrid: only its attention layers are made bidirectional, and the Gated DeltaNet layers stay causal. This is a base model, with no instruction tuning. Paper: dQwen3.5: Hybrid-Attention Diffusion Language Models. Code: https://github.com/AntonXue/dQwen Needs a CUDA GPU and transformers>=5.13 (tested with torch 2.7.1+cu128, flash-linear-attention 0.5.1). generate decodes the whole canvas at once, committing positions above a confidence threshold (tau=0.9); pass blocklength=32 for left-to-right block decoding, or tau=None, stepsperblock=k for a fixed budget. The 50B-token checkpoint from the paper is…

Open weights apache-2.0 9B parameters 262,144 tokens transformers

View model

Model · Text generation

nexgen-flash-v1-merged-averaged

Sampson Kontomah

basemodel: - Qwen/Qwen3.5-9B pipelinetag: text-generation libraryname: transformers

Open weights apache-2.0 9B parameters 262,144 tokens transformers

View model

Model · Text generation

WikiQwen-9B

Devon Yanitski

Ask anything, get a how-to article. WikiQwen-9B is Qwen3.5-9B fine-tuned to answer every message the same way: as a tidy, step-by-step how-to article in markdown, with a picture caption above each step. Hand those captions to WikiQwen-Illustrator and they become illustrations. Send it a question ("how do I keep basil alive?"), a problem ("my bike chain keeps falling off") or just "hi", and it sends back a guide. Every reply has the same shape: - a # How to … title and a short intro - one or more ## Method N: or ## Part N: sections, or a single ## Steps - an [IMAGE: caption] line above each step, written for the Illustrator to draw - steps written as N. Bold summary. Details…, with bullets…

Open weights cc-by-nc-sa-4.0 9B parameters 262,144 tokens transformers

View model

Model · Text generation

zkhapo3.5-9b

Zkhapo

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Open weights 9B parameters 262,144 tokens transformers

View model

Model · Text generation

JEV-9B-MLX-NVFP4

Nativ

MLX conversion of autotrust/JEV-9B for mlx-vlm, with the System 1 LoRA (including its lmhead update) merged and lmhead, the decision readout, kept unquantized. JEV is a decision model: it returns a probability for every option of a question in one forward pass and does not generate text. JEV support is not yet in an mlx-vlm release

Open weights apache-2.0 9B parameters 262,144 tokens mlx

View model

Model · Text generation

idefics-9b

HuggingFaceM4

How do I pronounce the model's name? Watch a Youtube tutorial IDEFICS (Image-aware Decoder Enhanced à la Flamingo with Interleaved Cross-attentionS) is an open-access reproduction of Flamingo, a closed-source visual language model developed by Deepmind. Like GPT-4, the multimodal model accepts arbitrary sequences of image and text inputs and produces text outputs. IDEFICS is built solely on publicly available data and models. The model can answer questions about images, describe visual contents, create stories grounded on multiple images, or simply behave as a pure language model without visual inputs. IDEFICS is on par with the original closed-source model on various image-text benchmarks…

Open weights other 8.9B parameters 2,048 tokens transformers

View model