Jev directory
CommunityJev Models & Reimplementations
S1 LLM Auto-Router
A CPU-first decision model from Decision Models that answers the seven questions an LLM auto-router needs (category, difficulty, stakes, four flags) in one ONNX pass; MIT weights plus a Jev-compatible hosted API.
- Reading time
- 2 min
- Published by
- Community
- Author
- Decision Models
- Last verified
- Oct 2026
Used inJev for Software & DevTools
Highlights
- Answers seven router questions per request, including needs_tools, needs_vision, needs_long_context and follow_up as yes/no probabilities.
- Built on jhu-clsp/mmBERT-small pruned to 11 layers and distilled from a fine-tuned mmBERT-base teacher, with per-head calibration temperatures.
- ONNX with int8 embeddings is 183 MB and needs neither PyTorch nor transformers: about 14 ms p50 on four pinned EPYC cores and 26 ms on four Xeon cores.
- On 2,175 held-out synthetic requests across 22 languages it reports 0.898 category accuracy, 0.819 all-seven-exact and 0.742 routing agreement.
- The card places it ahead of Jev 1.13, Winnow-12B and Weiche-395M on its routing test; MIT weights with a hosted API at $0.001 per million input tokens.
Quickstart
bash
pip install onnxruntime tokenizers numpy huggingface_hub
python example.py "Refactor this ABAP report to use CDS views"Watch out
Card reviewed 2026-10-17; all benchmark numbers are self-reported on the author's own held-out routing set, whose gold labels are a plurality vote of up to five LLMs, the independent blind set is only 100 requests with about plus or minus 8 points of noise, and the API pricing is a launch price that can change.
Questions about this entry
Short answers drawn from the linked project or page.
A decision model that classifies a request to an LLM before you route it. In one CPU pass it answers the seven questions an LLM auto-router needs to pick a model, returning typed answers rather than generated text.
Seven, with fixed option sets: category (nine choices), difficulty (five levels), stakes (four levels) and the yes/no flags needs_tools, needs_vision, needs_long_context and follow_up. It answers only those.
It is served as a Jev-compatible decision model on the Decision Models API, model id s1-llm-auto-router at /v1/systemone, and the card compares it head-to-head with Jev 1.13, which it places behind on this routing task.
Just a CPU. The weights are ONNX with int8 embeddings at 183 MB and need neither PyTorch nor transformers; the card reports about 14 ms p50 and 47 ms p99 on four pinned EPYC 9555 cores and 26 ms on a Xeon.
On 2,175 synthetic requests in 22 languages: 0.898 category accuracy, 0.974 mean flag accuracy, 0.819 all-seven-exact and 0.742 routing agreement, versus Jev 1.13 at 0.845 and 0.630 routing; gold labels come from LLM votes.
The held-out set is synthetic and the gold labels are a plurality vote of up to five open LLMs, not human labels. On a separate 100-request blind set the scores drop to 0.88 category and 0.60 routing, with about plus or minus 8 points of uncertainty.
The weights are MIT. Hosted, it costs USD or EUR 0.001 per million input tokens in both tiers, with output free, as listed on 15 October 2026; the card notes this launch price can change.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Hacker News: ASTGL replay chose local Laya routing over hosted Jev on one Mac Studio setup.
Jmeg8r15 pointsDoes an open-weight decision model beat a hosted one? Jev vs. LayaRead the thread on Hacker NewsReddit: A walkthrough of routing between models with Jev.
X: Standard One SH (StandardThinking)
Facebook: Intake Router keeps diagnosis out of the answer space at the front desk.
FacebookClinicnotes
More Jev resources like this
The v3 Jev-Style decision model at 2B in GGUF: F16 3.78 GB and Q8_0 2.01 GB match PyTorch FP32 on 100% and 99.7% of parity rows, 73.6% on JevBench v1.4.1 public items.
Jev Models & Reimplementations#community#open-models#local
Community
A local yes/no decision model that speaks Jev's wire format on a laptop CPU, refusing Choice and Score until those question types ship.
Jev Models & Reimplementations#community#python#open-models
Community
feder-cr/jevC++
feder-cr/jevC++A Python server in front of llama-server that reads one-token label probabilities and exposes them as POST /v1/systemone.
Jev Models & Reimplementations#community#python#open-models
Community
khimaros/verdictPython
khimaros/verdictPythonA LoRA plus 24-slot decision head over Qwen3.8-27B from Ming Xu's train-jev project: it serves TypeSafe's System One API with 0.851 Kev transfer-v4 accuracy and ECE 0.053, and needs an 80 GB GPU.
Jev Models & Reimplementations#community#open-models#local
Community
A 24 GB-GPU alternative from the train-jev project: Qwen3.5-4B plus a rank-16 LoRA and 24-slot decision head scoring 0.647 PK accuracy, 40.4 Decision Index and 43 ms p50 latency, ahead of Kev-4B at the same size.
Jev Models & Reimplementations#community#open-models#local
Community
A small distilled sibling of Perplexity's pplx-decider-v1.2: a Qwen3.5-4B decider that reads a state plus typed choice, noul and score questions in one pass, keeping most of the 27B model's Decision Index.
Jev Models & Reimplementations#community#open-models#multimodal
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.