Find the Right AI Model

Latest across the site

The strongest recent records, newest first — projects, posts, recordings and threads from every model we follow. A row gets here once somebody other than us has looked at it: a post with a hundred likes, a repository with a hundred stars, a thread with fifteen points. Each card links to our own page where we have one and to the source where we do not — a card marked source leaves this site.

• Jev · dev.to · source

  I got Jev to zero mistakes. I'm still using Flash-Lite.

  An eleven-minute field report whose conclusion is in the title: the author got Jev to zero mistakes and still ships Gemini Flash-Lite. This is the useful shape of a negative result — not 'the decision model is bad', but 'here is the job where the cheap general model was already enough'.

  15 reactions2026-10-08

• Jev · Hacker News · source

  Jev-Driven SRE Diagnosis: What Worked and What Failed

  A research group handed Jev a view of a cluster and let its fast judgments choose where to investigate and which evidence supported a diagnosis, then published all 105 diagnoses including the ones that went wrong. Negative results with the failure cases printed are rare in this topic, and worth more than another demo.

  50 points2026-10-07

• Jev · Hacker News · source

  Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers

  Red Hat's AI Safety team benchmarked nine guardrail candidates across four methodologies — CPU-scale pre-trained classifiers, LLM-as-a-judge and decision models — to ask whether Jev-style models earn their place for risk detection. It is the first enterprise-scale comparison we have seen on this question, and its answer leans toward the cheap old thing rather than either vendor's story.

  166 points2026-10-02

• Jev · Hacker News · source

  How accurately calibrated is Jev?

  A measured attack on the calibration claim, with the whole experiment series run for under $4. The author's description of the architecture — a pre-trained transformer with a classifier bolted on the end — is the clearest short explanation of what Jev is that we have read from outside TypeSafe, whether or not the conclusion holds.

  55 points2026-10-02

• Jev · Hacker News · source

  Jeeves. Reasoning improves Jev-like decision models

  PostHog's reasoning Jev-style classifier: a 9B model with a diffusion drafter, trained with SFT and CISPO, MIT-licensed with weights on Hugging Face. Its title carries the claim worth testing — that adding reasoning improves a decision model, which runs against the argument that speed is the whole point.

  242 points2026-09-29

• Jev · Hacker News · source

  Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms

  An 0.8B Jev-compatible decision model family you can train at home, shipped with GGUF builds, adapters and a hosted playground at jeffhub.ai. It is the most-discussed item in this column — 231 comments — and most of that argument is about whether a model this small holds up on real classification.

  576 points2026-09-28

• Jev · Repository

  dzhng/jevgrep

  A CLI that finds code by asking what it does, built for coding agents: Jev picks the files and the source context worth reading instead of the agent grepping its way in. The README's headline is a cost claim — "Same intelligence. About 30% lower cost." — measured on their own ten-task SWE benchmark.

  2,479 stars—

• Jev · Hacker News · source

  A single function Jev-like wrapper for LLMs, including vision models

  A short argument with code: wrap any LLM — vision models included — in a single function that returns a decision rather than text. 135 points and 12 comments, and a useful counterweight to the framework-shaped projects in the builds column, because it shows how little scaffolding the idea actually needs.

  135 points2026-09-26

• Jev · Hacker News · source

  Turning GLM-5.3-Flash into a Jev-like decision model

  PrivateMode's write-up of getting Jev-like behaviour out of a standard open model by crafting the prompt so the first output token is the answer: one forward pass, then a benchmark against both Jev and Laya. One of very few head-to-head measurements in this column that was not produced by a vendor.

  72 points2026-09-26

• Jev · Hacker News · source

  Ollaya – Ollama for open-source, Jev-style decision models

  The week's largest discussion of the decision-model stack: an Ollama-style daemon for open decision models behind TypeSafe's wire format. 587 points and 29 comments when we read it, and the argument in the comments is about whether a registry is the missing piece or another layer nobody needs.

  587 points2026-09-25

• Jev · Hacker News · source

  Show HN: Jev Plays Pokémon Red

  A Show HN that answers the first question anybody asks about a model this fast: can it play something harder than Tetris? Pokémon Red, open-sourced, with the author's own caveat that Jev is not yet fast enough for Doom. The comments are mostly about frame budgets and where the loop breaks.

  262 points2026-09-25

• Jev · Hacker News · source

  What Is RLCD? The Secret Behind Jev

  An explainer for the training method behind this class of model, with the objection attached: the top comment points out that the piece never defines the acronym in its own title. Useful as a pointer to the technique, not as the reference for it.

  60 points2026-09-24

Both sections, in full: Jev (209) · Laya (79)