Jev directory
CommunityJev Models & Reimplementations
StartLux-Decision v1 27B Q8_0 GGUF
The Q8_0 8-bit GGUF build of StartLux-Decision-v1-27B, 28.60 GB for llama.cpp, served through a decision wrapper that the card measures at 100% same decisions as the original and 209 of 231 on JevBench public.
- Reading time
- 2 min
- Published by
- Community
- Author
- StartLux Labs
- Last verified
- Oct 2026
- Languages
- Shell
- Capabilities
- Self-hostedRuns locally
Highlights
- Q8_0 build of StartLux-Decision-v1-27B: 28.60 GB and 100.0% same decisions as the original on JevBench public.
- Answers 209 of 231 JevBench public items (90.5%); Q4_K_M agrees on 96.5% of items and BF16 on 100.0%.
- llama-server serves the weights; startlux_decision.gguf_server adds the prompt format, letter readout and calibration.
- Requires llama.cpp build b10454 or newer; -ngl 99 offloads every layer to the GPU, leave it out on CPU machines.
- Plain chat gives no decisions: the option-letter readout and per-type temperatures live in gguf_server.py.
Quickstart
bash
hf download startlux-models/StartLux-Decision-v1-27B-Q8_0-GGUF --local-dir StartLux-Decision-v1-27B-Q8_0-GGUF
cd StartLux-Decision-v1-27B-Q8_0-GGUF
pip install -r requirements.txt
llama-server -m StartLux-Decision-27B-Q8_0.gguf -ngl 99 -c 16384 --parallel 4 --port 8081 &
python -m startlux_decision.gguf_server --model-dir . --llama http://127.0.0.1:8081 --port 8090Watch out
Card reviewed 2026-10-17; the agreement percentages, JevBench counts and file sizes are the publisher's own measurements against the original weights through the startlux_decision package. The weights are CC BY-NC 4.0 even though the inference code is Apache-2.0, and llama.cpp chat alone gives no decisions.
Questions about this entry
Short answers drawn from the linked project or page.
The Q8_0 8-bit GGUF conversion of StartLux-Decision-v1-27B, 28.60 GB for llama.cpp. The decision server in the repository turns it into the same typed decisions, with a probability for every option, as the original model.
The card measures 100.0% same decisions as the original weights across the 231 public JevBench items, answering 209 correctly (90.5%). Q4_K_M is the smallest file at 16.55 GB with 96.5% agreement.
Q8_0 (28.60 GB, recommended), Q4_K_M (16.55 GB, smallest) and BF16 (53.81 GB, the original weights unchanged); Q8_0 and Q4_K_M are standard llama.cpp quantizations of BF16 without an importance matrix.
Download the repo, pip install -r requirements.txt, start llama-server with the Q8_0 file (-ngl 99 offloads every layer to the GPU) and then run python -m startlux_decision.gguf_server pointed at the llama server; llama.cpp build b10454 or newer is required.
No. Plain chat does not give decisions: the option-letter readout and the per-type temperatures live in gguf_server.py, not in the weights, and the file holds the text decoder only.
CC BY-NC 4.0: free for research and other non-commercial use with attribution, and commercial use requires a separate licence from StartLux Labs. The inference code in startlux_decision/ is Apache-2.0.
The TypeSafe /v1/systemone format: POST a state and typed questions such as a choice with criteria or a yes/no, and receive a probability for every option. The example sends a billing ticket asking which team should handle it.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Standard One SH (StandardThinking)
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsMore Jev resources like this
The Q8_0 8-bit GGUF build of StartLux-Decision-v1-9B, 9.53 GB for llama.cpp, with the card reporting 100.0% same decisions as the original weights and 200 of 231 correct (86.6%) on JevBench public.
Jev Models & Reimplementations#community#open-models#quantized
Community
The Ollama build of iapp's OpenThai-SystemOne: a 0.8B Thai and English decision model whose v0.3 weights are fine-tuned for Ollama's /v1/systemone letter readout, reporting a 74.3 macro across 13 public subsets for Q8_0.
Jev Models & Reimplementations#community#open-models#local
Community
A 27B decision model from StartLux Labs that takes text, JSON or images up to 262,144 tokens and returns a probability per option through the TypeSafe /v1/systemone format, reporting Decision Index 0.2.1 63.88 and 208 of 231 on JevBench public.
Jev Models & Reimplementations#community#open-models#local
Community
A 27B decision model from torchcast-ai that answers choice, noul and score questions together in one forward pass over the TypeSafe-compatible /v1/systemone format; the card reports Decision Index 0.2.1 of 65.00, +7.09 over Jev 1.13, at 33 ms per yes/no question.
Jev Models & Reimplementations#community#open-models#local
Community
A local daemon that pulls open decision models and serves them on TypeSafe's /v1/systemone, so a Jev client switches hosts with TYPESAFE_BASE_URL.
Jev Models & Reimplementations#community#rust#open-models
Community
ollaya-dev/ollayaRust
ollaya-dev/ollayaRustA community Q8_0 GGUF of Mapika/decider-0.8b, the smallest System One typed-decision model: one forward pass returns calibrated Choice, Score and noul answers, reporting 0.405 on JevBench hard-public at 920 MB resident.
Jev Models & Reimplementations#community#open-models#local
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.