WRITING

Decision models, in depth.

Articles on System One decision models — how they differ from LLMs, the Laya and Jev architectures, calibration, and publishing to the registry.

October 1, 2026 · 2 min read

Add your System One Models certificate to LinkedIn, and let anyone check it

How to put your certificate of completion on your LinkedIn profile in one click, and how a recruiter or a colleague checks that it is real: the certificate's page, its number and its QR code.

Read the article

October 1, 2026 · 2 min

Learn System One models in an hour: a free course with a certificate

Getting started with System One models is a free course in nine short lessons, from what a decision model is to publishing your own. Finish it and claim a certificate with your name that anyone can check.

coursecertificatesystem-one-models

September 30, 2026 · 3 min

OpenAI, Liquid AI, Upstage and 17 more: the new System One models

Twenty decision models and APIs joined the registry this week: OpenAI's Decisions API, Liquid AI's d1, Upstage's Solar Decide, two more hosted APIs, and fifteen open models from InternLM, Juspay and others.

system-one-modelsjev-alternativeshosted-api

September 30, 2026 · 1 min

OpenDXP is now NoulXP

The open standard for System One models has a new name: NoulXP. Same format, same answers, same badge. Install it with pip install noulxp; everything made with OpenDXP still runs.

noulxpstandardrelease

September 30, 2026 · 4 min

What a decision costs: four decision models on a $0.51-an-hour GPU

We served Julia 1, Laya, Decider 2B and our package of AnyJev L0 (Nokia's method) on Qwen3-1.7B on one rented NVIDIA A40 and measured decisions per second, latency, cost per million decisions and how many models one GPU can hold. The cheapest cost $0.46 per million decisions, the most expensive $11.79.

benchmarkscostgpu

September 30, 2026 · 3 min

ConstraintViolationError exporting a transformer to ONNX with torch 2.8, and a six-line fix

torch.onnx.export(dynamo=True) with named torch.export.Dim objects fails its first two capture strategies on torch 2.8 and falls back to draft_export, four times slower. Naming the dimensions as strings fixes it.

onnxpytorchexport

September 30, 2026 · 4 min

Same model, different answers? Four decision models on a Mac and an NVIDIA GPU

We moved six decision-model packages from an Apple M4 to an NVIDIA A40 and an Intel Xeon, and checked every probability against answers recorded on the Mac. Several things changed the answers, none of them raised an error, and two settings fixed most of it.

noulxponnxruntimellama.cpp

September 30, 2026 · 3 min

Q4_K_M changed 14 of 90 decisions and made nothing faster: quantizing a decision model

We quantized our package of AnyJev L0 (Nokia's method) on Qwen3-1.7B to Q8_0 and Q4_K_M and checked every answer against the model's own. On an NVIDIA A40, 4-bit flipped 14 of 90 decisions, 8-bit flipped 2, and neither was faster.

quantizationggufllama.cpp

September 30, 2026 · 3 min

Removing option-order bias costs k times the compute. A prompt cache gives half of it back

Asking a language model a multiple-choice question once per rotation of the options removes its position bias. On an NVIDIA A40, a prompt cache cut the cost of those rotations by 2x, decoding only a third of the tokens, with every decision unchanged.

llmmultiple-choiceposition-bias

September 30, 2026 · 4 min

Six models on one GPU: how one slow model starved the rest, and the fix

Six decision models on one NVIDIA A40: two causal models' prompts made an 8 ms model wait 129 ms. A thread per model, with four computing at once, raised the total from 45 to 80 decisions per second at 16 clients per model, and from 44 to 69 at one.

gpuservingmulti-model

September 30, 2026 · 4 min

Give your AI agent a calibrated classifier: a decision model as an MCP tool

Agents make many small decisions, and asking an LLM for each one is slow and gives no usable confidence. Here's how to give Claude, Cursor or any MCP client an open decision model as a tool that answers in a fraction of a second, with a probability for every option.

mcpagentsclaude

September 30, 2026 · 4 min

Your classifier says 96% and is right 72% of the time. 250 labels fix that.

Most decision models we tested were surer than they were right. Three numbers per model, fitted to 250 labels, brought one from 96% confidence down to its real 72% without touching its weights. What you label with, and which data you use, matter as much as how many.

calibrationtemperature-scalingclassification

September 30, 2026 · 4 min

Batching made our CPU slower, and our GPU slower until we sorted by length

Batching is supposed to be free throughput. For decision models it slowed an Apple M4's CPU by up to 2.2x. On an NVIDIA A40 it paid off once each pass held only rows of similar length, and then the same GPU served 1.6x more.

batchinggpuonnxruntime

September 27, 2026 · 4 min

What is an AI decision model?

A model that returns an answer from a set you define, with a calibrated probability, instead of writing text. How it works, how it differs from an LLM, a classifier and JSON mode, and when to use one.

decision-modelssystem-oneexplainer

September 27, 2026 · 5 min

Open-source Jev alternatives, and how to choose one

Jev's weights are not published. Within ten days of its release, more than twenty open System One models appeared that you can run and fine-tune yourself, and eleven of them answer Jev's own API. What each one is, and how to pick.

jevsystem-oneopen-weights

September 23, 2026 · 3 min

Why decision models need their own registry

A general model hub can store a decision model perfectly well. Storage was never the problem — the metadata, the evaluation and the workflow are.

registryinfrastructuresystem-one

September 22, 2026 · 3 min

What is a System One model?

A System One model returns a decision, not a paragraph. That single difference changes how you evaluate it, deploy it and reason about it.

system-onedecision-modelsfundamentals

September 22, 2026 · 3 min

NoulXP 0.2: Nokia's AnyJev runs the same on any machine

NoulXP 0.2 adds an HTTP API, an MCP binding for AI agents, and rotations. The first model asked the new way, Nokia's AnyJev on Qwen3-1.7B, makes the same decision as Nokia's own code on all 90 test questions.

noulxpstandardanyjev

September 22, 2026 · 2 min

Tev1, explained — Together AI's $17 recipe for a Jev-style classifier

Tev1-4B-experimental is a LoRA fine-tune of Qwen3.5-4B that reads a state and up to 24 options and answers with one letter. Together published the model, the recipe and the tutorial; the weights licence is still being decided.

tev1together-aisystem-one

September 21, 2026 · 3 min

System One models vs LLMs — when a decision beats a paragraph

LLMs and decision models are not competitors. They answer different questions, and using the wrong one is expensive in a way that is easy to miss.

system-onellmcomparison

September 20, 2026 · 3 min

Publishing your first decision model

A walk through the whole loop — write a manifest, create a repository, publish a version, record the lineage — in about ten minutes.

guidepublishingsystemone-yaml

September 20, 2026 · 5 min

MLX vs PyTorch for fine-tuning Laya on an M4, measured

System One Studio now trains on PyTorch as well as MLX, so it runs on Windows and Linux too. We ran the same fine-tune on both on a 16 GB M4 — time, memory, accuracy, calibration, and whether the two checkpoints agree — and found a bug on the way.

layalaya-studiomlx

September 19, 2026 · 2 min

Laya, explained — an open decision model that runs on your own machine

Laya is an open-weights System One model built on a ModernBERT encoder. Here is how it scores options in one pass, what its checkpoints are for, and how to fine-tune it on a laptop.

layaarchitecturefine-tuning

September 19, 2026 · 3 min

The open reproductions — Nimble, Decider, Von, JevK5, OpenThai and the tiny ones

Within a week of Jev, individuals and small teams had rebuilt its shape from open bases. Some publish better calibration than the original. A guide to the ones worth knowing.

system-oneopen-weightsnimble

September 18, 2026 · 2 min

Jev, explained — the first System One model

Jev is the model that named the category. What it takes in, what it returns, what it costs, and what its makers have not said.

jevtypesafesystem-one

September 17, 2026 · 2 min

Kev, explained — a pointer head on a frozen Qwen

Kev is an open family of System One models from Jared Palmer that keeps a Qwen base frozen and learns only an adapter and a pointer head. Small to download, honest about calibration, and it speaks Jev's API.

kevsystem-oneqwen

September 16, 2026 · 3 min

Is there a Hugging Face for System One models?

Yes. Hugging Face hosts every kind of model; System One Models (systemonemodels.ai) is a registry built only for System One models, with the facts a decision model needs. What each does, side by side, and how to use them together.

system-one-modelsregistryhugging-face

September 15, 2026 · 2 min

GLiNER2.5-Decide, explained — labels you supply at runtime

Fastino's open-weight decision model does something the others do not. It classifies against label sets you hand it per request, extracts the spans that justify the answer, and enforces rules across decisions, all in one pass of a 340M encoder.

glinerfastinosystem-one

September 14, 2026 · 4 min

Every System One model, compared

Fifteen model families have appeared in ten days, from a closed hosted API to a 2.8 MB form-filler. What each one decides, what it costs to run, and what its makers actually measured.

system-onelandscapecomparison

September 13, 2026 · 2 min

CLM, explained — decisions as a contrastive lookup

Contrastive Language Models do not classify. They embed the state and every candidate action and pick by similarity, which is why a decision can cost 28 milliseconds and why the heads are 75 MB.

clmsystem-onecontrastive