Jev directory
CommunityJev Models & Reimplementations
Laya on ModelScope
The Laya family mirror on ModelScope: a 421M non-autoregressive System 1 decision model, Apache-2.0, with a pip-installable package, more than 100 languages claimed at about 33 ms per pass, and 16,713 downloads.
- Reading time
- 2 min
- Published by
- Community
- Author
- Convai Innovations
- Last verified
- Oct 2026
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- Non-autoregressive System 1 decision model that returns typed answers with mathematically calibrated probabilities in a single forward pass, about 33 ms.
- Claims 100+ languages and installs with pip install laya on Python 3.10 or newer.
- Optional extras: laya[serve] HTTP server, laya[mcp] MCP server, laya[langchain], laya[onnx], and laya[fast] GPU fast path.
- Trained with RLCD against strictly proper scoring rules; it never generates text, so there is nothing to parse and nothing to hallucinate.
- Repository holds three F16/F32 safetensors files: base model 842.6 MB, multilingual 643.8 MB, and typed-decisions 842.6 MB.
Watch out
Apache-2.0 per the metadata, but this repository sits under convaiinnovations, an organisation whose own description says it is a community mirror that hosts artifacts whose owners may not manage it, so it may not be the canonical Laya home. The long-document guidance is the project's own: 16 to 18 of 20 requests answered correctly up to about 4,000 tokens before the text, 8 to 17 beyond that.
Questions about this entry
Short answers drawn from the linked project or page.
A mirror of the Laya family: a multilingual, non-autoregressive System 1 decision model that takes a state and typed questions and returns typed answers with calibrated probabilities in a single forward pass, about 33 ms, across 100+ languages claimed.
pip install laya on Python 3.10 or newer. Optional extras cover an HTTP server (laya[serve]), an MCP server (laya[mcp]), LangChain and LangGraph (laya[langchain]), ONNX Runtime (laya[onnx]), and a TileLang GPU fast path (laya[fast]).
It trains with RLCD, reinforcement learning against strictly proper scoring rules, so reporting honest probabilities is the only way to maximize reward. Its Ollama sibling page reports an in-task calibration error (ECE) of 0.060.
Three safetensors files in F16 and F32: the base model at 842.6 MB, a multilingual build at 643.8 MB, and a typed-decisions build at 842.6 MB, with a model_size metadata figure of 421,293,830 bytes.
laya-multilingual reads up to 8,192 tokens with max_len=8192, and its default 1,024-token limit cuts long documents off. The project's own benchmark answered 16 to 18 of 20 requests correctly with up to about 4,000 tokens of text and 8 to 17 of 20 beyond that.
No. It never generates text, so there is nothing to parse and nothing to hallucinate; it scores options and returns probabilities. Its Ollama sibling page describes three question types: pick from a list, true or false, and rubric placement.
The ModelScope upload sits under the convaiinnovations organisation, which describes itself as a community mirror created to host open artifacts and invites owners to claim management rights; the readme links the project's own GitHub at github.com/NandhaKishorM/laya.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Both Jev 1.13 and 1.14 score 29 of 30 on 30 Turkish buyer and seller messages
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 NewsMore Jev resources like this
Cloudflare's 9B multimodal decision model on Ollama: text, JSON, or images in one non-autoregressive forward pass, Apache-2.0, and the lowest measured latency of any decision model in Cloudflare's Decision Index runs at 38.8 ms median.
Jev Models & Reimplementations#community#open-models#local
Community
Convai Innovations' 421M non-autoregressive decision model on Ollama: fine-tuned from ModernBERT-large, Apache-2.0, answering one question in under 10 ms on an M5 Max with 0.060 in-task calibration error.
Jev Models & Reimplementations#community#open-models#local
Community
Bespoke Labs' 9B decision model on Ollama, fine-tuned from Qwen3.5-9B: Apache-2.0, up to 64 typed questions per call, answers in under 100 ms on a MacBook Pro with an M5 Max, and 75.7% mean accuracy across 13 human-labelled datasets.
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/ollayaRustAn Apache-2.0 decision engine with a TypeSafe-compatible endpoint, document answers with source passages, and a ViZDoom demo that picks actions from structured game state.
Jev Models & Reimplementations#community#python#local
Community
deepanwadhwa/OpenDecisionPython
deepanwadhwa/OpenDecisionPythonA pointer-network decision model from Keyush Nisar that reads one agent state document plus typed questions and returns calibrated tool-routing probabilities in a single forward pass, reporting 89.00% zero-shot tool routing.
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.