Jev directory
CommunityJev Models & Reimplementations
PPLX Decider v1.2
Perplexity's 27B decision model rebuilt on Qwen3.8-27B with a joint request format that answers a whole request's questions in one sequence, reporting 68.09 on Decision Index 0.3 and 77.43 on vision.
- Reading time
- 2 min
- Published by
- Community
- Author
- Perplexity AI
- Last verified
- Oct 2026
- Languages
- Python
- Capabilities
- Open sourceSelf-hostedRuns locallyRequires API key
Highlights
- Reads a whole request in one sequence (state, then every question), so several questions about one state cost about as much as one.
- Self-reported Decision Index 0.3 overall 68.09 with joint requests, up from 62.30 for pplx-decider-v1.1-27b; vision 77.43 versus 73.22.
- Training adds decontaminated official training splits of some Decision Index benchmarks plus other decision data, combined by model souping.
- Handles text and images up to 1.6 MP; questions can inform each other, though grouping helps some benchmarks and hurts others.
- Needs Python 3.12+, a CUDA GPU with room for about 52 GiB of BF16 weights, and an access token for the private repository.
Quickstart
python
import sys
from pathlib import Path
from huggingface_hub import snapshot_download
checkpoint = Path(snapshot_download("perplexity-ai/pplx-decider-v1.2"))
sys.path.insert(0, str(checkpoint / "source" / "src"))
from pplx_decider import Decider
decider = Decider.from_pretrained(checkpoint, device="cuda")
response = decider.decide({
"state": "My integration keeps failing with a 401 since we rotated our API key. Please help.",
"questions": {"urgent": {"type": "noul", "instructions": "Is the customer blocked right now?"}},
})
print(response["answers"]["urgent"]["noul"])Watch out
Card reviewed 2026-10-17; the Decision Index and vision figures are self-reported on Perplexity's own public evaluation panel, including the deliberate use of official benchmark training splits that the card itself frames as a fairness argument, and the repository is private and needs a token despite Apache-2.0 metadata.
Questions about this entry
Short answers drawn from the linked project or page.
Perplexity's 27B System 1 decision model, built on the same Qwen3.8-27B backbone as v1.1. It answers typed choice, noul and score questions over text and images and returns probabilities and confidence instead of generated text.
Two things: a joint request format that reads the state and all questions in one sequence with a readout per question, and a reworked training mix that includes decontaminated official training splits of some Decision Index benchmarks.
All of a request's questions share one sequence: the state first, then every question, each with its own readout at a marker token after its options. Questions can inform each other, and several questions then cost about as much as one.
No. On the nine benchmarks whose requests carry several questions the two modes differ: grouping helps when the questions are facets of one decision and hurts when they are competing candidates. On the other 28 benchmarks scores are identical.
Self-reported on Decision Index 0.3: 68.09 overall with joint requests (and 67.75 one question per request) versus 62.30 for v1.1, with vision at 77.43 versus 73.22. All numbers are Perplexity's own.
Python 3.12+ and a CUDA GPU with room for about 52 GiB of BF16 weights plus working memory. The repository is private, so authenticate with a Hugging Face token that has access before downloading.
pplx-decider-flash-v1.2 is the same format and API at about one seventh the size, keeping most of the 27B scores: its card reports 60.05 with one question per request on the Decision Index versus 67.75 here.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
X: Standard One SH (StandardThinking)
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
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
A 4B decision model taking text, JSON, or images up to 262,144 tokens: 52.75 on Decision Index 0.2.1, 204 of 231 on JevBench public, and 26.0 ms for a three-question request on one H200.
Jev Models & Reimplementations#community#open-models#local
Community
InternLM's multimodal structured decision model, fine-tuned from Qwen/Qwen3.5-4B: a shared state, a schema of named questions and optional images go in, and every question returns an answer distribution from one forward pass. The card reports a 90.02 seven-suite average.
Jev Models & Reimplementations#community#open-models#multimodal
Community
InternLM's 0.8B multimodal structured decision model, fine-tuned from Qwen/Qwen3.5-0.8B: a state, named questions and optional images produce an answer distribution per question in one forward pass. Self-reported seven-suite average 79.38, Brier 0.530.
Jev Models & Reimplementations#community#open-models#multimodal
Community
InternLM's 2B multimodal structured decision model, fine-tuned from Qwen/Qwen3.5-2B: a state, named questions and optional images produce an answer distribution per question in one forward pass. Self-reported seven-suite average 84.68, Brier 0.437.
Jev Models & Reimplementations#community#open-models#multimodal
Community
Mohit Garg's Apache-2.0 vision-language decision model on Qwen/Qwen3.5-4B, the recommended default of the imajev family: it reads photos against records and text, returning calibrated probabilities and a trained can't tell.
Jev Models & Reimplementations#community#open-models#vision-language
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.