autotrust/JEV-35B-NVFP4 — new model trending #21 on Hugging Face
AutoTrust released JEV-35B-NVFP4, a 4-bit quantization of its 35B decision model that cuts GPU memory by 65%.
AutoTrust released JEV-35B-NVFP4, an NVFP4 quantization of its JEV-35B System 1 decision model, built on Qwen3.5-35B-A3B (35B parameters, about 3B active). The 10,240 routed expert MLPs are quantized while attention, routers, the vision tower, and the decision head stay bf16. Download size falls from 71.9 GB to 25.6 GB and vLLM weight memory from 66.5 GiB to 23.3 GiB. Decision Index 0.3 is 59.18 versus 59.75 for bf16, with 96.6% of text answers unchanged and similar single-request latency on one B200.
- NVFP4 experts cut weights from 71.9 GB to 25.6 GB.
- vLLM GPU memory falls from 66.5 GiB to 23.3 GiB.
- Decision Index drops 0.57 points to 59.18; 96.6% of text answers match bf16.
- Vision score falls 0.85 points; single-request B200 latency stays similar.
Full article2,921 words · extracted from huggingface.co · click to collapse
# autotrust/JEV-35B-NVFP4
[](https://news.guru.so/ui)
## 🔴 Live: AutoTrust/GuruSearch recommendation & search demo → [news.guru.so](https://news.guru.so/ui)
> [!IMPORTANT]
> **Try AutoTrust/GuruSearch live (new, 11 October 2026).** A recommendation and search demo in which every ranking is a
> calibrated System 1 decision, made in about half a second:
> * **Recommendation:** the latest headlines from 6 news feeds, ranked by importance.
> * **Search:** results from Google News, Bing News and Yahoo News re-ranked by relevance, side by side with the
> search engines' own order, plus a short answer with citations.
### [autotrust/JEV-35B](https://huggingface.co/autotrust/JEV-35B) with NVFP4 routed experts: the same model in 23 GiB of GPU memory instead of 66 GiB
## NVFP4 version
This repository is **autotrust/JEV-35B with its 10,240 routed expert MLPs (40 layers × 256 experts) quantized to
NVFP4**. Everything else is the bf16 model unchanged: attention, linear attention, the shared expert, the routers, the
vision tower, the MTP layer, the lm_head, the System 1 adapter, the decision head and the temperatures. Below the NVFP4
comparison, the rest of this card is the JEV-35B card; its tables are the bf16 results unless marked otherwise.
| | JEV-35B (bf16) | **JEV-35B-NVFP4** |
|---|---:|---:|
| download (weights) | 71.9 GB | **25.6 GB** |
| GPU memory for the weights (vLLM) | 66.5 GiB | **23.3 GiB** (−65 %) |
| Decision Index 0.3, public suite (our scoring with the kit) | 59.75 | 59.18 |
| same top-1 answer as bf16 (308,559 Decision Index answers) | — | 96.6 % |
| approximate public vision score (our rebuild, 13,361 rows) | 72.08 | 71.23 |
| same answer as bf16 (vision rows) | — | 95.2 % |
| median single-request latency, text / image (one B200) | 215 / 378 ms | 227 / 371 ms |
| computer use, numbered boxes + element text (60 tasks) | 95 % | 95 % |
| computer use, numbered boxes only (60 tasks) | 38 % | 37 % |
| robot arm, pick and place (20 scenes) | 75 % | 65 % |
* **Text: −0.57 on the Decision Index.** All five areas within 1.2 skill points of bf16 (Knowledge 0.422, Language 0.628,
Retrieval 0.706, Tools 0.753, Arts 0.413); the largest per-benchmark changes are Home appliance simulator −5.7,