Picture published by the creator on Hugging Face.
JEV-35B: Rank news and search results with a decision model
Added · Updated
Made by AutoTrust AI Lab
View profile on Hugging Face (opens in a new tab)JEV-35B evaluates up to 256 options in one pass.
Developers building ranking systems can explore its news demo, which orders headlines and compares search results.
Model details
Original model autotrust/JEV-35B
Input Text Output Labels
Ways to use it Browser demo · Run locally
One reviewed local configuration. Add My computer details to compare them.
BF16 vLLM on NVIDIA · vLLM · The pinned card documents vLLM on one GPU and notes 80 GB or more for this BF16 checkpoint. This BuildTube threshold is a separate file-size estimate at 4K context.
How this configuration was checked
96 GB VRAM · reviewed estimate Source lists 71.9 GB (file-size). Estimate uses 71.91 GB exact BF16 checkpoint bytes, 0.084 GB 4K KV cache (10 full-attention layers of 40 hybrid layers × 2 KV heads × 256 head width × K/V × 2-byte values), 4 GB runtime buffers, and 20.006 GB remaining in the 96 GB VRAM tier.
BF16 · 4,096 token context · source context 262,144 tokens · batch size 1 · concurrency 1 · Single-user 4K comparison; the source vLLM command supports a much longer context.
Source checked 2026-10-21 · revision a5d0d580bb9324f0f4c07dcba5c5ca91e813e7c9. pinned model card documents BF16, vLLM single-GPU setup, 80 GB source recommendation, and the 262,144-token context; estimate uses exact checkpoint shard bytes and config
Role in this family: Built from another model.
Qwen/Qwen3.5-35B-A3B (opens in a new tab)What it does
JEV-35B is a model for evaluating choices, including ranking items by relevance or importance. It builds on Qwen3.5-35B-A3B and can consider up to 256 options in one pass. Its published news demo shows how that capability can support recommendation and search tools.
How to use it
For a first look, open the AutoTrust/GuruSearch demo. Enter a news search and compare the model's reordered results with the original order from Google News, Bing News, and Yahoo News. The demo also provides a short answer with citations. Its recommendation view ranks headlines collected from six news feeds.
Common use cases
- Compare model-ranked news search results with the search engines' original ordering.
- Browse headlines from six feeds ordered by importance.
Can I use this?
Requirements Use the published demo to explore the ranking behavior. Hosting the standard model involves a 72 GB download and 66.5 GiB of GPU memory for weights alone. A complete hosting procedure is not documented.
Caveat AutoTrust reports its own benchmark scores and a median request time measured on one B200 GPU. Those figures describe that evaluation setup. JEV-35B has no separate thinking mode.
Worth knowing The reported scores and latency are AutoTrust's measurements. The standard model's weights require 66.5 GiB of GPU memory. Licensed under Apache License 2.0.
Why it's here
Not trending this week
Last seen on the Hugging Face trending list on 15 October 2026. It stays on BuildTube so you can still find it.
- 2,009 people have liked it on Hugging Face.
- It was downloaded 1,661,130 times in the last 30 days.
Numbers checked on 8 November 2026; not refreshed since.
Behind it
View model pageBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?
Quick questions
- What is JEV-35B?
- JEV-35B is a model for evaluating choices, including ranking items by relevance or importance.
- What can I use JEV-35B for?
- Compare model-ranked news search results with the search engines' original ordering; Browse headlines from six feeds ordered by importance
- Can I run JEV-35B on my computer?
- It runs on your own hardware with a model runtime. Hosting the standard model involves a 72 GB download and 66.5 GiB of GPU memory for weights alone. AutoTrust reports its own benchmark scores and a median request time measured on one B200 GPU.
- What do I need to use JEV-35B?
- Model hosting needs substantial GPU memory; the published news demo offers a first look.
- What license does JEV-35B use?
- Licensed under Apache License 2.0.
More from this week
- kimi-k3-in-c Run a 2.78-trillion-parameter model on 8 GB of RAM
Get text answers from the huge Kimi K3 AI model using one CPU and no graphics card.
- P-H3-Image-Comfy Make and edit 8-megapixel images in ComfyUI
Turn short text requests into 8-megapixel images and edit pictures on your own graphics card.
- Idol Turn one picture of an object into a 3D Gaussian splat
Idol turns one photo, render or generated image of an object into a 3D Gaussian splat file.
Explore by topic
Post this
Download share imageFor makers
Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.
README badge (Markdown)
[](https://buildtube.app/p/hf-autotrust-jev-35b/)Embed card (HTML)
<iframe src="https://buildtube.app/embed/hf-autotrust-jev-35b/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="JEV-35B on BuildTube"></iframe>