Source: vectorize-io/hindsight ↗Hindsight uses hs-decide-1-flash to prune and deduplicate recall, check whether evidence answers a request, consolidate new facts and decide whether a memory could change a knowledge page. It first tried this with Jev. In the team's own Hindsight 1.2.0 evaluation, glm-5.3-flash plus the decision model scores 87.1 at $7.72 per thousand cases, against 89.0 at $147 for gemini-3.8-flash alone. Reflect quality falls with some LLMs. gemini-3.1-flash-lite judges all runs and also appears in the comparison.
Vectorize · ★ 58,477 · MIT · hs-decide-1-flash
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Launch and evaluation → Source: MIT licence →
Source: browser-use/jev-ultrafast ↗Zurich to London on Google Flights in 7.1 seconds, including model calls, generated text, browser work and loading waits. Swapping to a decision model took median task time from 9.450s to 7.092s and median browser protocol calls from 1,092 to 101.
One natural-language goal, on the real web, at 1x speed. Maker-reported: three repeats of one task on one browser profile.
- 7.09 smedian task, down from 9.45 s
- 101browser calls, down from 1,092
browser-use · ★ 23,396 · MIT · Jev
Figures from the project's own README and docs/performance.md. The maintainers note this is three repeats of one task on one browser profile, not a general benchmark.
Source: GitHub → Source: Measurements →
Source: dzhng/jevgrep ↗A CLI that uses Jev to find source files and context for coding agents. Its author-run ten-task study reports 25.8% lower estimated total task cost with 8/10 solves in both conditions. A later change cut Jev cost about 59%, but raised combined cost 2% to 3% against that saved cohort, again with 8/10 solves.
David Zhang · ★ 3,199 · MIT · Jev
Catch-up September experiments, discovered through an in-window push. The ten Python tasks were repeatedly studied; saved baselines were not rerun. Costs include failed attempts and use list-price estimates, not reconciled invoices. Nineteen empty responses leave some later usage unknown; the report gives a ceiling.
Source: GitHub → Source: Licence → Source: Total-cost run → Source: Later combined-cost study → Source: Provider code →
Source: jarrodwatts/jev-trader ↗Attempts one trade decision per Monad block on the Kuru MON-USDC market, and holds when the model misses the block deadline.
Jarrod Watts · ★ 3,050 · MIT · Jev
Source: Liquid4All/cookbook ↗A pixel-art racer where d1 picks left, centre or right on every tick, with a mode that races it against Jev. Liquid AI's own example, in their cookbook repo, so the star count is for the whole repo.
Liquid AI · ★ 2,623 · no licence · d1
Source: awlevin/typesafe-computer-use ↗OCRs the screen, then classifies the next action. The claimed per-step cost is the maintainer's own figure, not independently measured.
awlevin · ★ 1,251 · MIT · Jev
Source: luxonis/oak-examples ↗YOLO-E locates objects and a local d3-lite Q8_0 model answers up to four typed questions per crop. The standalone RVC4 app keeps live video separate from the background decision worker and shows the input crop beside its answer. Luxonis publishes a warehouse demo and reports on-device tests, without matched accuracy or crowd-throughput measurements. Requires Luxonis OS 1.40 or newer.
Luxonis · ★ 983 · Apache-2.0 application code; model and dependency terms separate · d3-lite
Vendor demo and on-device execution report; not independently reproduced
Source: Application licence → Source: Runtime → Source: Demo and requirements → Source: On-device report →
Source: devagrawal09/jev-review ↗A staged review workflow with a local dashboard: each stage is a bounded judgement rather than a written critique.
Dev Agrawal · ★ 686 · MIT · Jev
Source: fhshaik/typesafe-mario ↗An agent that plays Super Mario Bros. by classifying the next action from structured emulator state rather than generating a plan.
fhshaik · ★ 456 · no licence · Jev
Source: Sidiora-Labs/codify ↗A local agent workflow tool written in C. Its recap command sends each session transcript to Solar Decide, which labels every statement as a goal, decision, constraint, done or noise before a chat model writes the summary.
Sidiora Labs · ★ 433 · MIT · Solar Decide
Source: kitze/unclutter ↗A browser extension that decides, element by element, what on a page is clutter, with reusable template rules.
Kitze · ★ 382 · MIT · Jev
Source: RomanSlack/jev-drone ↗A drone experiment in MuJoCo with a small judgement model in the control loop, on camera input. The maintainer calls the harder tunnel run a partial result.
Roman Slack · ★ 282 · MIT · Jev
Source: agent-labs-dev/fastbrowse ↗An LLM plans, reads and writes; Jev picks each action from the controls indexed on the page, and code stops irreversible clicks without authorization. On 46 tasks run three times each against the hosted Browser Use agent, it passed 99.2% of runs to 96.3% and cost $0.0296 a run to $0.3180, 10.7x less, but was slower: 18.9 s a task to 15.3 s (published campaign release-0.7.1).
agent-labs-dev · ★ 155 · MIT · Jev
Source: ChetasLua/jevmeter ↗Scores every sentence of a video and renders the running result as an overlay.
ChetasLua · ★ 108 · MIT · Jev
Source: mars-tw/jev-trading-command-center ↗A paper-trading research app where Jev makes the decisions, wrapped in an animated 3D command center with model pets and a ten-timeframe research panel. Every trade is simulated; there is no real order path.
mars-tw · ★ 63 · MIT · Jev
Source: uu889/laya-opencv ↗A local inspection workbench: OpenCV measures the image, hard rules decide anything numeric, and a Laya decision model cross-checks the result. Built for fruit picking, weed and pest monitoring, and steel pipe and textile checks. Version 3.0 also trains your own Laya-format decision models.
uu889 · ★ 58 · MIT · Laya
Source: nineninesix-ai/tyger-s2s ↗A browser voice agent where the TYGER decision model is the ears: it transcribes you and decides when you've finished, when you've cut in, and when to say "mm-hm", while an LLM writes the replies.
NineNineSix · ★ 53 · Apache-2.0 · TYGER
Source: levantolabs/sagecut-release ↗QA for Remotion videos: code measures every frame, the Sage decision model answers yes/no questions about each one, and only the frames it flags go on to the slower critics.
Levanto Labs · ★ 47 · MIT · Sage
Source: devanshbatham/commit-miner ↗Classifies git diffs and messages into bug fixes, security fixes with CWEs, and change types.
Devansh Batham · ★ 39 · no licence · Jev
Source: scs0209/laya-guard ↗Checks coding-agent tool calls with rules and a local multilingual Laya fine-tune, returning allow, review or block. VENDOR-REPORTED, base to fine-tune: the model alone held 32 then 39 of 43 attacks and allowed 25 then 44 of 50 everyday commands. Six end-to-end verdicts improved and two worsened, missing a preset no-regression criterion. Accuracy on 162 held-out training rows rose from 0.611 to 0.852. The author inspected the benchmark and evaluation sets between training runs, so they are less independent than a fresh test.
scs0209 · ★ 35 · MIT · Laya
Source: Fine-tune and evaluation →
Source: OmniJev/Flide ↗An inference engine that shares state processing across question branches and serves /v1/systemone. In the email demo checked on 12 Nov 2026, the team reports eight questions per email with OneJev1.6-0.8B in bfloat16 on one H200, and 28 seconds for 10,000 emails from 64 clients. This is a throughput example, not a spam-accuracy result.
Flide team · ★ 35 · Apache-2.0 · Local decision models
Author-run records; no fresh model inference reproduced
Source: Licence → Source: Server → Source: Verification code → Source: Email demo and timing setup →
Source: backmeupplz/jev_antispam_bot ↗A Telegram bot that checks spam and links related messages before deleting them, with admin exemptions and a keep-on-failure policy. Its published evaluation records 75 live Jev calls, including two rejected prompt changes. Those calls did not exercise Telegram deletion. The README says its production classifier is now a Laya fine-tune distilled from Jev.
Nikita Kolmogorov · ★ 21 · MIT · Jev, Laya
Licence, implementation and recorded live model calls checked 6 Nov 2026; no fresh Telegram run.
Source: Licence file → Source: Integration code → Source: Recorded model calls →
Source: markrussinovich/venn-vdm ↗A decision layer over reusable embeddings and recorded outcomes. Its download-free CPU quickstart uses authored vectors; text input needs a separate frozen reader and trained head. Code, an arena runner and a recorded-run manifest are public. The head has separate redistribution permission, and the docs warn that probabilities are not calibrated across tasks.
Mark Russinovich · ★ 17 · MIT code; trained head has separate redistribution terms · Venn
Maker's code and recorded runs · VENDOR-REPORTED; not independently reproduced
Source: Licence → Source: Recorded arena run → Source: Head and reader terms →
Source: vercel-labs/jev-ai-sdk-form-router ↗Routes lead, contact and issue forms with Jev through Vercel AI Gateway and the AI SDK. It accepts Jev's destination when confidence is at least 95%; otherwise, or if Jev fails, openai/gpt-6-luna-fast evaluates the submission independently. The result shows the destination, deciding model, timings and an email preview, with optional delivery.
vercel-labs · ★ 15 · MIT · Jev
Source: yorkeccak/model-vs-market ↗A live site where OpenAI Decisions, Jev and Cloudflare's Clef put a probability on live questions after reading recent news, never the odds. Their probabilities are compared against Polymarket and Kalshi prices.
yorkeccak · ★ 12 · MIT · Decisions API, Jev, Clef
Source: nouli-ai/nouli ↗Decision models forget everything after each answer; Nouli keeps a session. Type facts as you learn them and start a line with ? to ask a new question; earlier facts and answers stay in the history. A Rust CLI and library that can chain sessions into decision graphs, with Jev through its TypeSafe adapter or other providers.
nouli-ai · ★ 12 · MIT · Jev
Source: Javihaus/zeropoint ↗Scores option labels and an implicit none state using a bundled ONNX encoder and fitted head. The maker publishes saved comparisons, model hashes and fitting code. Canonical option sorting and one-case inference are part of its consistency recipe. The 900-case matched comparison reports 53.6% correct including none, with half the cases having no right option. Short input budgets and truncation flags remain explicit.
Javier Marín · ★ 12 · Apache-2.0 · ZeroPoint
Source: Licence → Source: Saved run tables →
Source: berget-ai/guardrails-md ↗A small decision model scores every bash command against your GUARDRAILS.md in about 100 ms (maker's figure), before an opencode, pi or Claude Code agent runs it. Commands scoring above the threshold for data loss, secret leaks or rule breaks are blocked; the README warns it can still let risky ones through.
Berget AI · ★ 11 · MIT · berget/bev (System One-compatible)
Source: mertkayacs/jevalt ↗Local System One server and CLI for Deem, Karar and Wähler, with optional unknown answers and short reasoning. The maker publishes 390 live requests across the three languages, including misses. Emberwick plays recorded model decisions in a browser village; those replays are not fresh inference.
Mert Kaya · ★ 7 · Apache-2.0 · JevAlt
Source: Licence → Source: Decision engine → Source: Recorded live requests →
Source: Civalski/claujev ↗A Claude Code plugin that asks Jev short typed questions while Claude works: how big is this request, is this command destructive, was the work really done. It suggests an effort level per request, and in enforce mode a confident check can block a "done!" when the work looks incomplete.
Civalski · ★ 4 · MIT · Jev
Source: seanparkross/slack-thread-police ↗A self-hosted Slack bot that privately suggests moving a channel message into a thread. Jev checks the recent message window by default; a local System One server can replace it. A person chooses whether to move or ignore the message. Debug logging can include Slack payload text.
Sean Park-Ross / Parklife, Inc. · ★ 4 · MIT · Jev
README screenshots and integration code; not independently run
Source: Licence → Source: Detection code → Source: Privacy and model settings →
Source: mraad/duckdb-jev-spatial ↗A native DuckDB extension that adds seven Jev-style SQL functions backed by a local Laya model: DuckDB handles geometry and distance, Laya reads the text fields. It uses no hosted model or API keys.
mraad · ★ 3 · Apache-2.0 · Laya
Source: philrenaud/practical-data ↗A Chrome extension asks a decision model which HTML table columns to shade and which end is better. Code parses values and paints the colours. The default free provider forwards table text to Jev; users can supply their own provider key or use its Cloudflare backend. The README reports colour-or-leave-plain matching 284 of 285 labelled columns on 26 pages. Labels and thresholds were tuned on those pages, so this is a regression check. The replay uses recorded model answers and does not retest the model.
Phil Renaud · ★ 3 · MIT · Jev
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: MIT licence →
Source: arslanrakh/perfect-recall ↗A Hermes memory provider, Python library and MCP server that asks remote Jev to match stored memories against an agent’s criteria. A frozen predecessor scores 85.8% final-answer accuracy versus Mnemosyne’s 48.3% on a custom 120-question LongMemEval-S test. The renamed release has not had a fresh independent quality test; eligible memory text leaves the machine.
arslanrakh · ★ 3 · MIT · Jev
Published author-run records, checked 5 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: MindrLabs/typed-decision-slay-the-spire ↗Runs Bespoke Nimble-9B and three Laya checkpoints as players over the same 200 seeds of a headless Slay the Spire simulator, zero-shot and fine-tuned. The results are the authors' own.
MindrLabs · ★ 2 · MIT · Nimble, Laya
Source: diegoag64/jev-devtools ↗A drop-in React inspector that turns an app's Jev requests into interactive diagrams of the context, each question and the returned answers. Marked experimental.
diegoag64 · ★ 2 · MIT · Jev
Source: BerriAI/better-backlog ↗A local web page and CLI that reads a repo's GitHub issues and lets you filter them by typing a criterion, such as people blocked from upgrading. Each criterion is asked of a decision model through your TypeSafe key or a gateway; add another to narrow the list.
BerriAI · ★ 2 · MIT · Jev
Source: sushrutkhandelwal/unorphan ↗A local app you point at a messy folder like Downloads: it works out which of your projects each loose file belongs to, reading the README or CLAUDE.md in your code folders, and files it into a matching folder under your chosen output location. Nothing moves until you press Move, and one click puts it all back.
sushrutkhandelwal · ★ 2 · no licence · Jev
Source: rchatham/SystemOne.swift ↗Native Swift and MLX inference for local d1 text, images and speech, with a shared HTTP server and an older von backend. The author publishes Python reference vectors and layer-by-layer parity measurements. Head-logit differences are 5.8e-5 absolute for d1 and 6.0e-5 for von. These are implementation checks, without a task-accuracy benchmark. The HTTP wire accepts text states; native Swift calls carry media.
Reid Chatham · ★ 2 · Apache-2.0; SystemOneD1 under LFM Open License v1.0 · d1, von
Author-run reference vectors and source checked 6 Nov 2026; not rerun locally.
Source: Licence file → Source: d1 terms → Source: Integration code → Source: Reference vectors → Source: Port verification →
Source: opper-ai/jev-vs-kev ↗Benchmark code and a 362-question test set that sends each item to Jev and to Kev 4B, and reports accuracy, calibration, latency and cost.
Opper · ★ 1 · Apache-2.0 · Kev, Jev
Source: sunnydachs/span01-eval ↗Tests Span-01 Lite as a second check on whether narration text is in the right language, after a regex pass, against a general chat model. The author notes the 23-case test set is small.
sunnydachs · ★ 1 · MIT · Span-01
Source: patrikcze/mail-triage ↗Classifies Apple Mail messages as for you, to review or junk with a Laya model running on your own computer, and shows a report of what it would do. Mail never goes to an external AI service, and it never deletes anything.
patrikcze · ★ 1 · Apache-2.0 · Laya
Source: GuitarWag/linkedin-focus-filter ↗A Chrome extension that hides LinkedIn posts that don't match your stated interests, using Kev 0.6B running in the browser on WebGPU. Post text never leaves the browser.
GuitarWag · ★ 1 · MIT · Kev
Source: CosmoIknosLab/PathJev ↗A decision model for histopathology: given a tissue image, a question and candidate answers chosen at run time, including 'none of these', it returns a probability for each in one forward pass. Built on MedGemma-1.5-4B with LoRA adapters and a decision head, with code to rebuild every number in the authors' preprint.
Cristian Tommasino and Antonio M. Rinaldi · ★ 1 · Apache-2.0 (code); weights for non-commercial research only (MedGemma HAI-DEF terms) · PathJev-P (own, MedGemma-1.5-4B base)
Source: Simonlee711/MedJev ↗Runs an open System One model locally on Apple silicon against ten medical benchmarks, from USMLE-style questions to emergency-department admission and 30-day readmission, and compares it with zero-shot LLMs and supervised baselines.
Simonlee711 · ★ 1 · MIT (code); model weights CC BY-NC 4.0 · OpenJev Flash 9B
Source: prgrmrwy/dsh-auto-model-routing ↗A DeepSeek Harness plugin where Jev, via OpenRouter, judges each message as simple, standard, complex or escalated and raises or lowers only the reasoning effort of the model you picked. If Jev fails, it switches itself off.
prgrmrwy · ★ 1 · MIT · Jev
Source: scalefreegan/jev-ants ↗Each ant is a Jev agent that sees the cells around it and its neighbours' headings and decides which way to move and whether to settle. A thousand small decisions turn a random scatter into a word or an uploaded picture; the maker puts a word at about fifteen seconds and seven cents.
scalefreegan · ★ 1 · MIT · Jev
Source: bit-cook/Jev-X-Blocker ↗A Chrome extension that hides spam and noise in X posts and replies by combining Jev's probabilities for rules you write in plain language with local signals, and decides locally what to hide. Everything hidden can be restored in place, and if the API fails, posts are shown.
bit-cook · ★ 1 · MIT · Jev
Source: andletenkov/decisium ↗Write test steps as plain-English Gherkin and Jev picks the element, action and check for each one from options Decisium reads off the page; code, not the model, decides whether a step passed. Resolved steps are recorded so CI can replay them with no model calls.
andletenkov · ★ 1 · MIT · Jev
Source: parasite6/noullama ↗A desktop app that finds and downloads decision-model GGUFs from Hugging Face, runs them on llama.cpp, and serves a local API with choice, yes/no and score questions, plus a playground for trying presets.
parasite6 · ★ 1 · GPL-3.0 · Local GGUF models
Source: decision-tune/realstop ↗A Claude Code Stop hook with a local fine-tune of DecisionTune 1.0 that can block a turn ending that hands work back to the user. VENDOR-REPORTED on 864 test turn ends: the moderate setting caught 67.5% (195 of 289) of hand-backs at 15.8% false interventions, against 56.7% and 14.6% for phrase rules. That test split guided development. Both members were correct in only 16% (37 of 229) of synthetic pairs; the model did not beat phrase rules there. Warm model latency was a median 27 ms on an M5 Pro. The hook cannot verify claimed task completion.
decision-tune · ★ 1 · Apache-2.0 · DecisionTune 1.0 (RealStop fine-tune)
Source: Model and evaluation →
Source: HayateV30/luar ↗Labels spreadsheet text locally with Laya and exports answers, confidence values and review flags. On 300 texts per dataset, the author reports 93% English news-topic accuracy, 68% Portuguese review sentiment accuracy and 33% exact star-rating accuracy. The review labels come from customer ratings, which can disagree with the text.
HayateV30 · ★ 1 · MIT · Laya
Published author-run records, checked 5 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: ritza-co/hn-jev-sorter ↗A Chrome extension that asks Jev to sort Hacker News comments into reader-defined buckets, retaining the parent comment as context. The author’s demo records a 131-comment thread being classified and sorted. Six requests run concurrently, and an Other bucket catches comments outside the chosen categories. The recording shows the workflow; it does not measure label accuracy.
Ritza · ★ 1 · MIT · Jev
Licence, integration code and author-recorded demo checked 6 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Recorded demo →
Source: moudrkat/kneejerk ↗Moves llama.cpp decision serving code into a C library. Its recorded d1 request takes 678 ms on eight CPU threads. Saved CLI/server parity tests stay within 0.01 for Laya, exceed that probability tolerance for H2O and Clef, and reject the tested Perplexity package. On a Pixel 8a in Termux, the maker reports a 464 ms yes/no decision with d1-2B Q4_K_M on four threads and 2.1 GB RSS. Without repacking, that question takes 1,210 ms at 1.1 GB RSS. iOS and Capacitor app bindings remain plans.
Kateřina Fajmanová · ★ 1 · MIT · d1, Laya, H2O-Lightning-4B, Clef
Source: Licence → Source: Saved parity output → Source: Phone measurement and build notes →
Source: jinyoung/nimble-pilot ↗A macOS app whose browser agent makes every choice, such as which element to click, as one classification by Nimble-9B running locally in Ollama. Low-confidence calls go to the person.
jinyoung · ★ 0 · Apache-2.0 · Nimble
Source: indranildchandra/hybrid-intent-router ↗Routes support messages through regex, then a small classifier, then decision models, then an LLM as a last resort, all running locally. Laya answers the guardrail and department questions, and Tev1 is the default for the next tier.
Indranil Chandra · ★ 0 · MIT · Laya, Tev1
Source: bidouilles/pacman-systemone ↗A one-file browser Pacman where the open directions at each step become the options in a single question to a local Tev1 model through Ollama's decision endpoint.
bidouilles · ★ 0 · no licence · Tev1
Source: anujgupta9860/intent-router-openai-decisions ↗A proof of concept that classifies incoming requests with the Decisions API preview and hands them to worker agents. Its README warns that the preview's request format is not yet published.
anujgupta9860 · ★ 0 · no licence · Decisions API
Source: saembit/jevquery ↗A TypeScript library where Jev picks the table, columns, filter, order and limit from what the schema offers, so it can choose a column but never write a query. Drizzle, Prisma or a SQL builder turns the plan into a real, parameterised query.
saembit · ★ 0 · MIT · Jev
Source: Marcyvan1777/jev-arena ↗A 3D boxing game where Jev, GPT-4o-mini and Claude Haiku pick each move from the same state and legal actions, and a seeded rules engine resolves the hits. The maker calls the 56-bout results a pilot, not a ranking.
Marcyvan1777 · ★ 0 · no licence · Jev
Source: jeromtom/petsona-catalogue-ai ↗Tags each pet product's species, life stage and allergens from the store's own text, asking Jev or Laya one typed question per value: supported, contradicted, unknown or conflicting, with the source sentence attached. Merchants can approve, correct and lock tags, or let a validated policy accept them automatically. A hackathon entry for AIBoomi Startup Weekend Kochi, 23-24 Oct.
Jerom Palimattom Tom (Rexin Dynamics) · ★ 0 · MIT · Jev, Laya
Source: Monkeypox966/arxiv-jev-digest ↗A GitHub Actions job where Jev scores each new paper from your chosen arXiv categories against each of your research interests, code ranks them and keeps the top few, and only those go to DeepSeek for a summary emailed to you. The default follows humanoid robotics papers.
Monkeypox966 · ★ 0 · MIT · Jev
Source: n8peace/politiciansbyjev ↗Ask a question and Jev, via Vercel AI Gateway, gives each of 100 well-known American politicians a yes/no probability; the faces above the threshold float up out of a physics pile.
n8peace · ★ 0 · MIT (code); portraits under their own licences · Jev
Source: ByronDarlison/call-coach ↗A macOS app that transcribes both sides of a call on your Mac with Whisper while Jev ticks off each of your questions once it has been answered, notes which question you just asked, and flags surprising answers for a follow-up. The maker puts Jev's cost at about 1 to 2 US cents for a 45-minute interview.
Darlison Holdings · ★ 0 · MIT · Jev
Source: balajibankapur/textbook-voice-tutor ↗Answers a seven-year-old's spoken questions from one scanned Grade 2 textbook chapter. Jev decides whether each new question can be answered from the chapter and, with the optional verifier on, whether the drafted answer's claims are supported by the retrieved passages; common questions are matched to cached answers. On the maker's 14-question test, the decisions API cut average answer time from 4.1 to 1.7 seconds and the cost of each gate call about 25 times, against the chat wrapper.
balajibankapur · ★ 0 · MIT · Jev
Source: blurboy1985/ssoc-decision-trail ↗Paste a job ad and watch Jev walk Singapore's 2024 occupation classification level by level, keeping the three best paths at each step, to a five-digit code, drawn as an animated pixel-art crowd. Six requests, about 1.4 seconds, per the README.
blurboy1985 · ★ 0 · no licence · Jev
Source: manikv723-jpg/JevVoice ↗Speak several instructions in one run-on sentence and Jev picks each action from a menu of skills, and any app from those installed on your Mac, while you are still talking. Anything outside the menu goes to Claude for a proposed shell command, which waits for your approval unless you turn on trust mode.
manikv723-jpg · ★ 0 · no licence · Jev
Source: sarath/krestel ↗A Thunderbird extension that tags your email with decision models served by Ollama's /v1/systemone endpoint, such as Tev1 0.8B, Tev1 4B or Nimble, and keeps learning from the tags you add by hand.
sarath · ★ 0 · MIT · Tev1, Nimble
Source: dmonteroh/sys1_mcp ↗An MCP server that gives an AI assistant yes/no, choice and score tools backed by whichever decision model you configure: any /v1/systemone endpoint such as Jev or Clef, or Laya running on Apple silicon. Its tools never name a model family.
dmonteroh · ★ 0 · MIT · Laya, Jev-Style-0.8B (any /v1/systemone endpoint)
Source: Dasge97/the-village ↗A pixel-art village in the browser where Jev, via OpenRouter, chooses what each adult does for the next two hours, while fixed rules handle urgent needs, night-time sleep and children. The maker built it to test whether Jev's choices stay sensible and reflect each villager's personality.
Dasge97 · ★ 0 · MIT (code); CC0 1.0 (graphics and sounds) · Jev
Source: blakestone-x/buildornot ↗Describe a side project and get BUILD, MAYBE or DON'T BUILD with the reasons: Jev rates pain, reach and feasibility out of five and judges whether the GitHub and Hacker News results it finds already do the job. A run costs well under a cent.
blakestone-x · ★ 0 · MIT · Jev
Source: senthilsweb/jev-pii ↗A DuckDB function that asks Jev typed questions about each database column: is it personal data, which kind, and whether GDPR, CCPA-CPRA, HIPAA, LGPD, PIPEDA or DPDP apply. Whether to keep, mask, hash or drop it is decided by code, not the model, and a failed call sends the column to review. Also runs in the browser.
senthilsweb · ★ 0 · Apache-2.0 · Jev
Source: rickcrawford/jev-relevancy ↗Grades search results with Jev on a four-level scale, checked first against Amazon's human-graded shopping queries, and stores runs, results and judgments in DuckDB with the metrics computed in SQL. The author recommends checking it against at least 300 of your own labelled pairs before trusting it on another domain.
rickcrawford · ★ 0 · MIT · Jev
Source: fmegahed/decisions-space ↗A Hugging Face app built on the R package decisions: upload a set of images or texts, ask Jev one question about each, and get every answer with its probabilities plus the R code to repeat the run. Visitors need no key or account; runs are checked against estimated session and daily budgets.
fmegahed · ★ 0 · MIT · Jev
Source: KoheiTakada/QuietFlow ↗An Android app that can use Jev to sort notifications into five classes, from action required to hide, alongside local safety rules and corrections it learns from you. It starts in observation mode and never cancels notifications there; Jev is opt-in with your own key.
KoheiTakada · ★ 0 · MIT · Jev
Source: gianviterbo/wps-watch ↗A public monitor of the West Philippine Sea with live ship positions, Sentinel-1 radar passes and Global Fishing Watch events. Jev answers four questions about each item from 13 news feeds, and two of them, whether it serves a Philippine reader and whether it carries PRC state framing, decide whether it is shown or held back. Live at wps.gdgt.link.
gianviterbo (Gadget Pilipinas) · ★ 0 · Proprietary (contact Gadget Pilipinas for reuse) · Jev
Source: arndth/tetrisarena ↗Jev, SAP-RPT-1, LLMs and a 15-line heuristic play Tetris on the same pieces and candidate moves, with live cost tracking. In the maker's 50-piece race Jev and the heuristic both cleared 19 lines and the others 18, a tie within noise; the difference was speed and cost.
arndth · ★ 0 · MIT · Jev
Source: Daewooki/decision-agent-lab ↗Calls Jev, OpenAI's Decisions API, Upstage Solar Decide and local Ollama models through one interface and logs version, latency, tokens and cost. On picking the right one of 12 tools for 40 Korean and English requests, the maker measured Jev 1.13 at 40 of 40 and OpenAI Decisions at 0.90, against 0.85 for a generative model choosing directly.
Daewooki · ★ 0 · MIT · Jev, Decisions API, Solar Decide, Nimble
Source: shohei81/jev-tos-checker ↗A Chrome extension that splits the terms of service on the page into clauses and has Jev check each for eight kinds of one-sided term, from selling your data to forced arbitration, marking only clauses that go well beyond what major services do and showing each next to the original text. The maker says it is not legal advice.
shohei81 · ★ 0 · MIT · Jev
Source: tejred213/ci-triage ↗Jev reads each failed test and decides the cause: environment or timeout, flaky assertion, genuine regression or API contract mismatch. Code re-runs the build only when every failure would be fixed by re-running (three attempts in all by default), gives flaky tests a low-priority ticket as well, and files tickets for real failures with the test's owner.
tejred213 · ★ 0 · no licence · Jev
Source: koltyakov/decide ↗A Go command-line tool and MCP server that gives coding agents and shell scripts Jev's yes/no, choice, score and multi-question calls, with offline request previews and batch runs.
koltyakov · ★ 0 · MIT · Jev
Source: Mou1z/wayfinder ↗Visitors type what they need in their own words and Jev, via OpenRouter, picks the one page on the site that fits, choosing only from your list of pages. With no key or a slow reply it falls back to word matching.
Mou1z · ★ 0 · MIT · Jev
Source: prefill-vanishing-point/order-matcher ↗Matches messy restaurant orders such as "2 cs portobellos, 5# limes" to exact catalogue items with Jev and flags uncertain lines for a person to check. The maker reports 96% exact matches on held-out test orders.
prefill-vanishing-point · ★ 0 · no licence · Jev
Source: lesderid/maddy-jev-filter ↗Sorts incoming mail for the maddy mail server into IMAP folders: Jev answers questions such as whether a message is spam or phishing, and rules in a config file turn the answers into a folder. It also learns from where you have filed mail before.
lesderid · ★ 0 · MIT · Jev
Source: dallascrilley/intune-why ↗Reads Intune compliance data from Microsoft Graph, groups failing devices by the setting that fails, and ranks fixes by how many devices each makes fully compliant. For custom settings its rules don't cover, Jev can optionally pick the remediation path. You can try a live demo with a sample tenant.
dallascrilley · ★ 0 · MIT · Jev
Source: Aladincz/Jev-Content-Filter ↗A browser extension that asks Jev about every ad on the Bazoš classifieds site against criteria you write in plain language, and collapses the ones that don't match. Other marketplaces and auction sites are planned.
Aladincz · ★ 0 · MIT · Jev
Source: royvandewater/jevflip ↗A small command-line tool that asks Jev to flip a coin N times and reports the share of heads.
royvandewater · ★ 0 · no licence · Jev
Source: uint4096/hndelegate ↗Polls new Hacker News stories every two minutes, runs each unprocessed one through decision trees of Jev questions you define, and sends the stories that match any tree to a Telegram chat or channel.
uint4096 · ★ 0 · GPL-3.0 · Jev
Source: jflowers/opencode-auto-question ↗An opencode plugin for autonomous runs: when the agent stops to ask the human a question, Jev (or an LLM, if you set one; the default tries Jev first) picks from the offered options. Below a configurable confidence (0.6 by default) it gathers context in a side session and decides again; below 0.35 the human answers after all.
jflowers · ★ 0 · Apache-2.0 · Jev
Source: dantetekanem/pi-jev-lazy-load ↗A Pi extension that keeps most tool and skill descriptions out of the first request and loads them on demand. In its optional Jev mode, Jev decides before each prompt which ones the request needs.
dantetekanem · ★ 0 · MIT · Jev
Source: prashaantr/what-changed-its-mind ↗An open research essay on three competitive 2026 US Senate races: compare stored Jev forecasts across different sets of evidence and different orderings of candidates and sources, with the exact requests and saved responses to inspect. The author calls it an experiment, not a calibrated forecast.
prashaantr · ★ 0 · MIT · Jev
Source: iamvazghen/jevqueue ↗A Claude Code plugin that checks an agent's tool calls before they run, using hosted Jev when you configure a key and offline rules otherwise. In enforce mode it refuses a call with a reason; advisory mode only reports the verdict.
iamvazghen · ★ 0 · MIT · Jev
Source: brent-phillips/jev-for-marketers ↗A free guide, a Claude skill and four editable recipes for typed marketing-ops decisions: route this lead, tag this asset, is this reply interested. Jev makes the call; an LLM still writes the copy.
brent-phillips · ★ 0 · MIT · Jev
Source: Nixz0824/x-semantic-filter ↗A Chrome extension: describe what you don't want to see in plain words, and eligible posts on your X timeline get a probability per rule. By default, anything over your threshold folds into a one-line bar you can open again.
Nixz0824 · ★ 0 · MIT · Jev, Clef-omni
Source: ikeremerdem/laya-pet ↗A small ticket router: with the Laya backend selected, each support ticket goes to Laya, the open Jev-style model, with three questions (which team, how urgent, is it spam). The default backend is a keyword stub, with a slot left for Jev.
ikeremerdem · ★ 0 · no licence · Laya
Source: hckmstrrahul/parallel-ticker ↗A small demo with a live site: type a US or Indian stock and two Parallel Decisions calls, through Parallel's free MCP endpoint, assess its technicals and trading activity from the live web and return probabilities and sources.
hckmstrrahul · ★ 0 · no licence · Parallel Decisions API
Source: aaron-zhlg/odds-arena ↗Keeps the 50 busiest Polymarket questions forecast by several models, including Jev, Perplexity's Decider and an OpenAI model, all given filtered web search results but never the price. When markets resolve, each is scored on a leaderboard by Brier score, log loss and accuracy.
aaron-zhlg · ★ 0 · no licence · Jev, Decider
Source: freakngenius/jev-workspace ↗Give Jev a mission and some links: it gathers every item, judges each one live with counters and a running API bill on screen, and ends with a results page you can sort, spot-check and export. Runs on your own machine with your own keys.
freakngenius · ★ 0 · MIT · Jev
Source: nikspence/idec ↗A script that asks true/false and multiple-choice questions about an image, PDF, Word file, slide deck, spreadsheet or text, using a decision model on local Ollama. Validated with d1:3b and Clef-Flash; d1:3b is the default.
nikspence · ★ 0 · MIT · d1
Source: NoxLoveYa/space-sniffer ↗A Windows disk-space map that draws your drive as a 3D city and asks Jev which folders look safe to delete; those get a neon outline and land in a Reclaimable list, biggest first.
NoxLoveYa · ★ 0 · GPL-3.0-or-later · Jev
Source: byui-library/jev-email-triage ↗A step-by-step guide for people who don't write code: set up Jev with Claude Code and build a read-only assistant that ranks the last two weeks of Gmail into piles like Reply needed, Waiting on them and Possible phishing.
byui-library · ★ 0 · MIT · Jev
Source: Enayat-Hassani/exposure-aware-routing ↗Simulates a moderation queue hit by a wave of privacy-invading posts the existing router did not prioritise: Claude reads 20 of those posts and 20 ordinary ones and drafts one yes/no question, and a fine-tuned version of the open decision model Laya asks it of every new post so the matches jump to the urgent queue.
Enayat-Hassani · ★ 0 · MIT · Laya
Source: nace-ai/drex-duckdb ↗Nace's DuckDB extension asks typed questions about rows from SQL and returns probabilities as column values, for routing support tickets, classifying documents or scoring data you already query. Nace also publishes MIT-licensed connectors for Snowflake, PostgreSQL, ClickHouse and Apache Spark, including Databricks SQL. Text and JSON calls default to drex-v2.0; images need an image-capable model such as drex-max.
Nace AI · ★ 0 · MIT · Drex 2.0, Drex Max
Source: Snowflake → Source: PostgreSQL → Source: ClickHouse → Source: Apache Spark and Databricks SQL →
Source: kristopolous/comfyui-cameo ↗A ComfyUI workflow that adds you to an existing video. Liquid's d1-2B judges every shot (framing, whether there's room for one more person, where everyone stands) and casts the best ones, then image and video models put you in beside the people already there.
kristopolous · ★ 0 · MIT · d1
Source: kubrakrcn/kariyer-decision-bench ↗Written at Kariyer.net R&D to pull seniority, department and work type out of job ads. On 600 real ads with 7 questions each (4,200 decisions), Jev 1.13.0 scored 71.5% against 64.6% for the open tau-4b and 65.7% for a prompted 4B LLM. tau-4b was better calibrated (ECE 0.093 against 0.156), and it scored 95.7% on its own training data, which is why the author warns against benchmarking a model on the set it shipped with.
kubrakrcn · ★ 0 · Apache-2.0 · Jev, tau-4b
Source: adityavsb/narrative-audit-jev ↗A harness that checks claims from earnings-call transcripts and footnotes against tables in DuckDB, on 64 synthetic cases. Perplexity's pplx-decider-flash-v1.2 scored 85.94% at about 1.4 s a case; a rule baseline scored 68.75% but caught only 2 of 9 Q&A deflections, which the decider caught 9 of 9. A chat model led on accuracy at 90.62%, with confidences stacked at 0.99 to 1.00. OpenRouter's chat endpoint rejected the decider with HTTP 400, so it runs through the separate /api/alpha/decisions endpoint.
adityavsb · ★ 0 · MIT · Decider
Source: gak4u/layagrep ↗jevgrep finds code by asking Jev thousands of yes/no questions. layagrep records those questions on SWE-bench tasks and fine-tunes the open Laya to answer them locally, with no API key and no code leaving the machine. A smoke run on 400 held-out questions lifted file-navigation AUC from 0.79 to 0.94 (Jev 0.97), but declaration relevance stayed at 0.58 (Jev 1.00).
gak4u · ★ 0 · Apache-2.0 · Laya, Jev
Source: cafe3310/decision-cli ↗Reads the probabilities of the answer tokens from one forward pass of an ordinary local LLM, with no fine-tuning, and serves them through a CLI, batch mode and a Jev-like API on Apple Silicon. On the 231 public JevBench questions, gemma-4-26b-a4b scored 84.8% at about 170 ms a question and ling-3.0-tiny 69.7% at about 45 ms; most of the misses are on the hard set.
cafe3310 · ★ 0 · Apache-2.0 · Local MLX models
Source: WyrmKeep/decision-model-kernels ↗A lab notebook for speeding up one decision model, H2O Lightning 4B, served through llama.cpp on an RTX 3060 Ti. Each request takes 2.13 to 2.45 times its roofline floor. The first measured change, a llama.cpp pull request, cut latency by 1.94% with bit-for-bit identical answers; the bigger gains in its list of 15 targets are labelled predictions.
WyrmKeep · ★ 0 · MIT · H2O-Lightning-4B
Source: ch7flyer/astrbot_plugin_jev_damocles ↗An AstrBot plugin that decides whether a bot in a group chat should reply. It merges bursts of messages, separates conversation threads, then asks Jev four questions, batched into one request when the backend supports it (was the bot addressed, does it follow, should it reply, how urgent). For messages without a direct mention, quote-reply or poke, it uses a high and a low reply threshold alongside other checks, and scores in between can be reassessed later; those direct interactions can skip the gate. The README suggests a 3 to 7 day dry run first.
ch7flyer · ★ 0 · MIT · Jev
Source: pyxl-dev/jev-router ↗A Claude Code plugin that asks Jev one choice question per message, low, medium or high effort, and applies the answer when Jev is at least 60% sure. It changes effort, never the model, so the prompt cache survives. Subagents are never routed.
pyxl-dev · ★ 0 · MIT · Jev
Source: HangChi/JevRouter ↗An agent runtime where an LLM reads the task, one Jev call answers four typed questions (next action, risk, whether to confirm, whether to reason further) and a local policy engine decides what may run. On 53 versioned test cases it hit the expected action 92.5% of the time against 96.2% for an LLM baseline, at about 1.0 s against 4.0 s, and asked for confirmation on 16 of the 17 cases that needed it.
HangChi · ★ 0 · MIT · Jev
Source: milesroxas/skill-tester ↗An eval harness for agent skills: headless agent runs, code checks, Jev judge packs and blind human review; the generic pack asks one yes/no question per skill rule. Graded with the writing pack, the reference skill's v2 run matched its original benchmark's report within 2 points on every layer and in total.
milesroxas · ★ 0 · MIT · Jev
Source: Alyosha28/dsh-cu-d1 ↗Uses Liquid's open d1-2B as a local System 1 for DeepSeek Harness Browser Use and macOS Computer Use, with approvals and routing. Downscaling wide screenshots before the model reads them cut one decision from 879 ms to 310 ms on a 2508-pixel window, and six consecutive trials all chose a click action. d1-2B itself is under the LFM Open License, not MIT.
Alyosha28 · ★ 0 · MIT · d1
Source: cars3/agastya ↗Retrieval that walks a document's table of contents, asking Jev one yes/no question per section, then ranks the sections it kept with EmbeddingGemma. On all 6,005 test queries of the public SearchTome release it returned 89.1% of gold sections at 3.6 sections per query, against 84.5% at 3 for EmbeddingGemma alone; its top-1 and top-3 recall equal the embedding model's. Jev alone, ranking flat, scored 54.2 top-1. It needs no training and runs on one 8GB consumer GPU, and failed upgrades are published too.
cars3 · ★ 0 · MIT · Jev
Source: uhiron55/physical-ai-2026-advanced-1-final-assignment ↗A course project where Jev chooses the transfer strategy for a CRANE+ arm in a Gazebo simulation, MoveIt 2 plans the path and ROS 2 controllers run it. In three episodes on authenticated Jev decisions with no fallback, Jev picked the home transfer at 0.94 to 0.96 confidence and all three placed the cube, 2.70 to 7.48 mm off target. These are scripted-teacher episodes; the README's separate learned-policy (ACT) tests are not reported as grasp success.
uhiron55 · ★ 0 · Apache-2.0 · Jev
Source: frederico-kluser/jev-droid ↗Drives an Android emulator from plain language: DeepSeek V4.1 Flash decides the next step, and Jev triages each screen (loading, error, permission prompt) and guards actions before they run. The three-step demo in the README, Settings to Network to Wi-Fi, cost $0.00019 and took 21 seconds; every step saves its screen, decision, tokens and cost.
frederico-kluser · ★ 0 · MIT · Jev
Source: tauseefau/topic-gate ↗Asks Jev one question per message, 300 to 600 input tokens, to tell whether the conversation is on the same topic, a subtopic or a new one, and recommends a session reset only when a new topic wins clearly (p of at least 0.65 and 0.25 ahead), so you stop paying to resend a dead transcript. Checked against live Jev on related, unrelated and short messages.
tauseefau · ★ 0 · MIT · Jev
Source: itsdula/laya-notes ↗A desktop notes app for Mac, Windows and Linux that keeps plain Markdown files and runs the open Laya model on your own computer. Each tag is a yes/no question, like "Is this note about a trip?"; Laya answers all of them for a note in one pass and keeps the tags it is confident about. Notes stay on the computer; the only network request is a one-time download of about 1.6 GB of model files.
itsdula · ★ 0 · MIT · Laya
Source: Paul-Yuchao-Dong/pi-omarchy-cu ↗Lets the Pi coding agent drive real Linux desktop apps on Omarchy (Hyprland Wayland) by reading the accessibility tree and asking Jev for the target element and the operation. It offers 11 Wayland-safe element operations. Jev-driven actions abstain below 0.55 confidence and need confirmation below 0.70, or 0.90 for destructive ones; direct operations on a known element run without Jev.
Paul-Yuchao-Dong · ★ 0 · MIT · Jev
Source: charles-forsyth/jev-decision-wall ↗Asks Jev 200 yes/no questions about one text in a single call and shows the answers as a heatmap. The keyless hosted demo replays recorded answers. VENDOR-REPORTED: about 185 ms for one question, 200 ms for 50, and 150 to 300 ms for 200; seven parallel calls returned 1,400 answers in 282 ms, against 1,092 ms back to back. The author measured Jev 1.14 from a US home connection and has not measured calibration.
charles-forsyth · ★ 0 · MIT · Jev
Source: AIAnytime/Jev-Router ↗An OpenAI-compatible router that filters models by hard constraints before Jev judges which one fits. VENDOR-REPORTED, on its own 30 prompts: tier accuracy was 100% against 53.3% for a heuristic, task-type accuracy 83.3% against 46.7%, and hard prompts sent to a weak model 0 against 14. Estimated savings against always using the strongest model, from token counts and configured prices rather than running the chosen models, were 35.6% against 68.7%; the heuristic's estimate was higher because it under-routed. Decision latency was 295 ms at p50 and 409 ms at p95.
AIAnytime · ★ 0 · MIT · Jev
Source: Ydjin1984/laya-zcode-stack ↗Runs multilingual Laya through a local Go/ONNX service, with a ZCode plugin for risk checks, context selection and 13 MCP tools. VENDOR-REPORTED on an RTX 3060, three runs per variant: total input fell from 359,085 to 239,544 tokens (33.3%) and requests from 16 to 9, but non-cached input rose 21.8%. Laya ranking alone cut total input 9.3% and non-cached input about 2.8%, within the observed range; the author has not established a fresh-token or cost saving. The /decide endpoint took about 26 ms at p50.
Ydjin1984 · ★ 0 · MIT · Laya
Source: Benchmark and caveats →
Source: House-elves/pr-wall ↗Puts open pull requests on a wall and asks Jev eight typed questions about each diff and its tests, using quarkusio/quarkus by default. It keeps the answers separate and never combines them into a score. Right/Wrong feedback records corrections and groups accuracy by confidence as you label answers; no accuracy figures are published yet. Its questions and Quarkus context are adapted from brunobat/quarkus-pr-wall.
House-elves · ★ 0 · Apache-2.0 · Jev
Source: heff0/Kev-claw-decisionmaking ↗Supplies OpenClaw with local typed decisions from LiquidAI/d1-Omni-2B-MLX-8bit on Apple Silicon. It has a native decision provider and a bridge for OpenClaw's TypeSafe integration. The repository name contains Kev, but the runner loads d1-Omni. VENDOR-REPORTED on an M4 Mac mini: about 200 ms per warm batch and about 3 GB of resident RAM.
heff0 · ★ 0 · MIT · d1
Source: nace-ai/inspect-drex ↗A scorer for the UK AI Security Institute's Inspect framework that asks Drex whether an output meets your criteria and keeps the probabilities, so you can change pass thresholds after the run. It defaults to drex-max, whose target changes with flagship releases; pin a model for repeatable runs. Drex Max accepts images, while drex-v2.0 accepts text and JSON.
nace-ai · ★ 0 · MIT · Drex 2.0, Drex Max
Source: raghavio/ownyourtimeline ↗Filters the For you and Following feeds in Chrome or Edge using a plain-language instruction and Jev. It sends one bulk TypeSafe request per timeline response, splitting to stay within input budgets, and removes posts scoring at least 0.90 before X renders them. Errors and timeouts leave posts visible; it checks at most 80 posts per response. The built extension has Chromium tests with X and TypeSafe responses stubbed. No filtering-accuracy figures are published.
Raghav Sharma · ★ 0 · MIT · Jev
Source: TechPreacher/jev-router-copilot ↗Jev sizes a Copilot message as tiny, everyday, large or hardest and sends it to a helper agent pinned to a model. Defaults are Haiku 5.5, Sonnet 5.5 and Opus 5.5, with the hardest helper configured to request high effort. Recorded helper runs used medium effort. The author records live runs of the hook in Copilot CLI 1.0.105 and VS Code 1.144, plus the CLI wrapper. The MCP fallback is tested offline only. Its 54 offline tests passed when checked on 03 Nov 2026; it publishes no savings benchmark.
Sascha Corti (TechPreacher) · ★ 0 · MIT · Jev
Source: Live CLI wrapper run → Source: Live VS Code run →
Source: wevans997-dev/clef-playa ↗Clef scores tiles from Jamen Percy's 2026 Burning Man gigapan, then the code sums expected RV counts, car counts and shade coverage. The author's 256 px run judged 8,785 tiles for $5.29, estimating 14,385 RVs, 18,084 cars and 256.8 acres of shade. These are estimates: vehicles under shade are hidden, and dense 512 px tiles undercount. Splitting validation tiles into 256 px quadrants brought RV estimates to 386 against 394 from the LLM counters.
wevans997-dev · ★ 0 · MIT · Clef
Author's recorded runs · VENDOR-REPORTED
Source: Xunzhuo/dsh-system-one-plugin ↗A DeepSeek Harness plugin uses vLLM-SR d3-ultra to triage steps, guard tool calls, strip injected lines and check whether work is done. This is the model maker's integration, by Xunzhuo Liu, co-founder of vLLM Semantic Router. In his small probes, the guard got 35 of 36 commands right, stopping all 16 risky ones, and the shield got 22 of 22 passages right. He reports a 0.49 s median across 53 guard decisions. Triage of 230 tickets took 121 s and 117,877 LLM prompt tokens, against 543 s and at least 510,986 for the LLM alone.
Xunzhuo Liu · ★ 0 · Apache-2.0 · vLLM-SR d3-ultra
Author's recorded runs · VENDOR-REPORTED
Source: Maker profile → Source: Maker's website →
Source: Jaydccq/mail-agent ↗Incoming Gmail goes through Jev triage before alerts reach Telegram or Lark. Claude or Codex CLI can draft replies, which wait for a Send tap. One Python process uses one SQLite file. The author's testing notes record real Jev calls on nine synthetic emails at about 0.1 s each, plus an end-to-end run with real Gmail, Telegram, Jev and Codex CLI; they report no triage accuracy figure.
Jaydccq · ★ 0 · Apache-2.0 · Jev
Author's recorded runs · VENDOR-REPORTED
Source: Testing notes →
Source: Potterluo/sluice ↗A coding harness lets a worker propose state transitions and a judge approve, reject or redirect them. Its 02 Nov 2026 real-model run used Jev 1.14 as judge and deepseek-v4-pro as worker. The author reports five judge calls for under $0.0003; the allowed-choice menu corrected three requests for nonexistent states. A separate five-task SWE-bench pilot does not identify its judge, so its score cannot be attributed to Jev.
Potterluo · ★ 0 · MIT · Jev
Author's recorded runs · VENDOR-REPORTED
Source: Ramolisdenneyous/Drone-Commander-Powered-by-Jev ↗In this local field test, Jev picks a unit's action from legal option IDs supplied by a tactics game's rules engine. The server checks the choice; a deterministic fallback handles late, malformed or weak replies. The author estimates about 75% lower token cost for those decisions with no obvious loss of tactical ability. Those are early playtest impressions. The first matches set both confidence and margin gates to zero, letting every legal Jev choice play.
Ramolisdenneyous · ★ 0 · MIT · Jev
Author's recorded runs · VENDOR-REPORTED
Source: elcann7/jev-issue-triage ↗A GitHub Action asks Jev nine questions per issue in one request, then a decision table turns the probabilities into labels and priority tiers. Probabilities between 0.3 and 0.7 count as unsure. It has a dry-run mode. The author's run passes 65 of 66 checks on 16 synthetic issues for about $0.001. That small synthetic set includes a fixture revised after its fake-example wording affected the answer.
Elcan Səfərli · ★ 0 · MIT · Jev
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: MIT licence →
Source: filip-a-9/jev-prompt ↗This Claude Code mod asks Jev through OpenRouter to grade the prompt being typed and name what it lacks. Hints are advisory; zero-data-retention routing is on by default. On 84 prompts, the band agrees with labels 71% of the time against an 80% target. Across 217 prompts with known outcomes, hints appeared on 25; 24% of those were followed by trouble (95% interval 11% to 43%), against a 10% overall rate. AUC is 0.60. Three blind Claude labellers supplied the labels. The harness is private, and whether the hints improve prompts has not been shown.
filip-a-9 · ★ 0 · MIT · Jev
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: MIT licence →
Source: code2saurabh/glassbox ↗Scans three Laya checkpoints for answers that ignore the state, option-order changes and negation failures. In the author's 04 Nov scan, 5 of 984 instructions collapse on English Laya, 5 on typed-decisions and 35 on multilingual. On a 216-answer realistic guardrail panel, 16, 6 and 42 answers respectively are wrong at confidence of at least 0.9. The panel uses six texts per word, and some scan questions are semantically odd. Results identify cases to check, rather than an overall model ranking.
code2saurabh · ★ 0 · Apache-2.0 · Laya
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Scan results → Source: Apache-2.0 licence →
Source: realizer-1966/layaForWeb ↗Runs Laya locally in a browser through ONNX Runtime Web, using WASM on CPU by default. The author reports 97.9% top-answer agreement with PyTorch across 48 questions for each quantized build. In Chromium, the default q8e8 build matches 11 of 12 top answers. A separate synthetic 108-case test of the original PyTorch model gets 59.3% of 312 answers right; that measures model behaviour, not browser parity. Quantized confidence values are approximate because calibration used the full-precision model.
YongSoo Park · ★ 0 · Apache-2.0 · Laya
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Synthetic test and limits → Source: Apache-2.0 licence →
Source: ElyasEbrahim/pgdocket ↗Stores versioned decisions in Postgres and checks calibration against the user's labels. The author's Phase 0 run used Jev 1.13 on 2,000 consumer complaints for about $0.30. Top answers changed on 4.8% of rows across three identical runs, and on about 16.5% when the same rows were rebatched. The repository includes measured Phase 1 ledger and routing runs but has no published release. Its claims about managed hosts have not been tested on Supabase, Neon or RDS themselves.
ElyasEbrahim · ★ 0 · Apache-2.0 · Jev / Laya
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Phase 0 findings → Source: Apache-2.0 licence →
Source: ErenTerakye/lowgate ↗Decision models choose one of 35 tilt-and-throttle manoeuvres on the same seeded lunar-lander courses. The game waits for the answer, so measured latency never changes the flight score. Across 160 levels per lane, the author reports 159 landings for a scripted autopilot, 156 for Jev, 151 for Kev-27B and 142 for the Laya fine-tune. Prompt selection and Laya training are documented; these results cover this simulator, with recordings available for replay.
Eren Terakye · ★ 0 · MIT · Jev, Kev, Laya, Clef-flash
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Recorded benchmark → Source: Prompts and training → Source: Licence →
Source: aryan-sentlyx/glimmer-system-one ↗Visual-inspection demo and benchmark for Muse Glimmer 30B with logistic decision heads. The maker reports 90.2% defect accuracy on 866 held-out images from known MVTec AD products. Half of each product's test images trained the head. Code, prediction records and a separate zero-shot path are public; the dataset is restricted to research use. The 21 ms early-exit measurement uses a truncated benchmark, while the supplied server runs full depth.
Aryan Srivastava · ★ 0 · Apache-2.0 · Glimmer System One
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Benchmark report → Source: Runtime and benchmarks → Source: Licence →
Source: demirelo/weight-edit-lab ↗Tests whether LLM-written weight edits or Python overrides repair Laya's routing of Banking77 messages into five departments. In a two-seed pilot, the best code-patch mean is 78.1% accuracy, against 76.9% for the best weight-edit mean at a different effort setting and 70.3% unedited. Test-selected fine-tuning reaches 75.0%. Budgets differ beyond label counts, and isolation relies on paths and log scans. Stored logs and summaries support the reported runs; saved edited weights are omitted.
demirelo · ★ 0 · MIT · Laya
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Laya results → Source: Licence →
Source: craigraphics/jarvis ↗Handles exact voice commands with patterns, then asks Jev to choose an action for natural phrasing; an LLM handles uncertain cases and writes text. The author reports 69 measured commands: patterns handle 54%, and Jev the other 46% at 116 ms median. All eight questions and comments return unknown and run nothing. Its automated tests use no network or real sign-ins.
William Craig · ★ 0 · MIT · Jev
Authors' published runs · VENDOR-REPORTED; not independently reproduced
Source: Decision routing → Source: Licence →
Source: fm4238704/reply-router ↗Rules handle unsubscribe and automatic replies first; local Laya labels the rest, with uncertain messages sent to an LLM. The author reports 110 of 117 selected replies correct at a 0.80 threshold out of 182 real replies. Rules were written after reading that same set, and model-drafted labels remain under review. Fresh held-out performance is unknown.
fm4238704 · ★ 0 · MIT · Laya
Published author-run records, checked 5 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: james-tn/decision-model-contract-judge-eval ↗Tests Microsoft-Decision-1 and a frontier reasoning model as contract-review judges on 2,091 ContractNLI and 1,224 CUAD judgments. Decision-1 with rules scores 82.9% and 90.0% accuracy. It reads overlapping chunks while the frontier model reads whole documents; added rules and confidence definitions differ. This is one author’s field test.
james-tn · ★ 0 · MIT · Microsoft-Decision-1
Published author-run records, checked 5 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: realizer-1966/laya-workflows-ko ↗A Korean adaptation of layaForWorkflows, using the multilingual Laya checkpoint in the browser. It records model answers for 37 examples across seven workflows. At branch threshold 0.50, its replay routes 15 automatically, 17 to people and five to blocking. Wording was adjusted against these examples, so this is a demonstration set.
realizer-1966 · ★ 0 · Apache-2.0 · Laya
Published author-run records, checked 5 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: realizer-1966/laya-rag-judge-ko ↗Runs multilingual Laya and a multilingual MiniLM embedder in the browser on Korean fictional documents. Its Korean report gives 62.5% accuracy on 72 claims with the default two-option question and 28% final-outcome agreement on 32 routing questions. The retained English README describes upstream results, not these Korean runs.
realizer-1966 · ★ 0 · Apache-2.0 · Laya
Published author-run records, checked 5 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: cyriusweng/jev-tern ↗A browser agent in Tern’s visible WebKit tab: Jev 1.14 judges steps and GLM-5.3-Flash supplies typed values. The author records nine successful runs across three Wikipedia tasks. Median times are 1.6, 3.4 and 1.8 seconds. The same tasks shaped several fixes; the browser-use comparison uses a different browser and model, so it does not isolate the decision model’s speed.
cyriusweng · ★ 0 · MIT · Jev
Published author-run records, checked 6 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published run →
Source: realizer-1966/layaForMemory-ko ↗A Korean browser adaptation of Laya Memory, using multilingual Laya and MiniLM to flag stale stored facts. On 62 hand-written cases, the author records 19.4% strict accuracy and 27.4% lenient accuracy with the default policy; it closes one of 22 stale facts and wrongly invalidates one standing fact. This sample also shaped the questions and thresholds. App-event pairs have not been recorded.
realizer-1966 · ★ 0 · Apache-2.0 · Laya
Published author-run records, checked 6 Nov 2026; not reproduced in this scan.
Source: Licence file → Source: Integration code → Source: Published Korean run → Source: Recorded answers →
Source: Who-Code/laya-browser-grounding ↗A browser element picker tested on an M2 Pro with 16 GB RAM. The author records 94 correct element picks from 94 live-site decisions across five runs, while only 91 of 95 browser steps finish. Page and action failures explain the difference. The wider grounding comparison finds no significant accuracy difference after multiple-comparison correction. Raw runs and a demo are public.
Who-Code · ★ 0 · MIT · Laya MLX
Author-run records and source checked 7 Nov 2026; not rerun locally.
Source: Licence file → Source: Grounding and browser code → Source: Recorded live run → Source: Results and limits →
Source: typestrict-lab/runtime ↗A Rust and ONNX HTTP server for typestrict decision models, with no model bundled in its image. A pinned TS-PRGuard download script checks file hashes; golden vectors record the browser reader’s real model outputs. HTTP tests check answer shapes and errors. The author’s Apple M5 strategy measurements compare latency, memory and throughput, not model accuracy.
typestrict-lab · ★ 0 · Apache-2.0 · TS-PRGuard
Author-run records and source checked 7 Nov 2026; not rerun locally.
Source: Licence file → Source: HTTP implementation → Source: Pinned test-model download → Source: Recorded model outputs → Source: HTTP contract checks →
Source: kkgzsprov/shakedown ↗A game playtest pipeline with planning between runs, model tactic selection and a code executor. Its recorded 36-run campaign made 247 Jev decisions, with 17 random fallbacks. All runs were valid, but the rule player had better outcomes: it escaped in all six aggressive runs while Jev’s aggressive player did not move. Three seeds and 12-second runs make this an exploratory demo.
kkgzsprov · ★ 0 · Apache-2.0 · Jev
Author-run records and source checked 7 Nov 2026; not rerun locally.
Source: Licence file → Source: Recorded campaign → Source: Replay-verified demo →
Source: Imperfective/smart-factory-ai-ops-center ↗A simulated factory gives Jev a menu of feasible actions and puts an independent safety gate before the virtual PLC. Stored real-model campaigns cover 1,260 shifts and 32,226 calls without API errors or fallbacks. Most plan-completion gain comes from the surrounding architecture; its deterministic planner matches the full system at lower operating loss. No real equipment is controlled.
- 1,260recorded real-Jev simulation shifts
Minkyu Kim · ★ 0 · MIT (code, documentation and simulation results); CC BY-NC-SA 4.0 (vision results) · Jev
Author-run simulation records, not a fresh reproduction. Confirmatory batches exclude mock runs and the standalone extra shift. The gain from adding Jev alone is not statistically significant.
Source: MIT licence → Source: Licence notice → Source: Jev client → Source: Stored runs →
Source: johnhenry/laya-js ↗Runs Laya through native MLX, WebGPU or a TypeScript CPU reference. The author records 756 of 756 matching top answers against Python across three checkpoints and four GPU configurations. This measures implementation agreement on 63 questions, without labelled task accuracy. Q4 changes one to five top answers per checkpoint on those questions.
- 756/756top answers matching Python across GPU configurations
John Henry · ★ 0 · MIT · Laya MLX
Author-run parity and performance records on a fanless Apple M2. Matching top answers does not mean identical probabilities; sustained thermal throttling changes timing.
Source: MIT licence → Source: Agent code → Source: Parity and timing records → Source: Recorded Python outputs →
Source: JuneYaooo/medical-jev-train ↗Training, inference and evaluation code for the medical-jev-4b adapter. The published GPU package check records 1,028 correct answers out of 1,514, with 100 over-limit questions counted wrong. Experimental tournament handling produces a different result, 1,050 correct.
JuneYao · ★ 0 · MIT · medical-jev-4b
Author-run records, not a fresh reproduction. The benchmark was reused during development; the original training file is lost and the public training pool is reconstructed. The model is uncalibrated and is not for clinical use or triage.
Source: GitHub → Source: Licence → Source: Inference code → Source: GPU package check → Source: Evaluation and limits →
Source: yangs1202/Kanana-2-v-decide-3b-2611-dgx-sparks ↗Pinned BF16 Docker recipe for Kanana on one NVIDIA GB10. Published memory, HTTP latency and Korean-exam records include actual choice, noul and score responses. The maker measured up to 17.12 GiB additional unified system memory under standalone load. A final image rebuild did not repeat full GPU loading; earlier loader runs and live-server smoke results remain separate.
yangs1202 · ★ 0 · MIT recipe; Kanana Open License weights and NVIDIA image terms · Kanana-2-v-decide
Source: Recipe licence → Source: Validation and scope → Source: Real three-type API response →
Source: oneryalcin/hunch ↗Versioned judgment specs with cached answers, human review, tests and a diff of decisions changed by a prompt edit. The bundled Jev command-guard result covers 38 commands and keeps a warning for a missed outside-project effect. Only its destruction question has gold labels; two other hazards do not. This is a catch-up project discovered through an in-window push.
Mehmet Öner Yalçın · ★ 0 · Apache-2.0; server directory Elastic License 2.0 · Jev
Author-run output; not independently reproduced
Source: Code licence → Source: Server licence → Source: Jev integration → Source: Recorded test output →
Source: Mohith26/jev-calibration-audit ↗Recomputes calibration and selective agreement from pinned, published WorkflowEvals outputs for nine models across four workflows. The saved results reproduce 36 headline scores. Its reference is a two-LLM probability consensus, not human truth; no fresh model is called. Case-level gates remain in-sample.
Mohith Gajjela · ★ 0 · MIT · Jev
Author-run records; no fresh model inference reproduced
Source: Licence → Source: Results → Source: Dataset matching → Source: Metric implementation →
Source: uran-lajci/system-one-multilingual-sentiment ↗Three-label sentiment code and saved predictions for Jev, Clef and Clef-flash on eight European languages. The zero-shot report covers 12,970 test items. Its comparison with five LLMs uses published paper scores with a different prompt format, rather than fresh matched calls. Broader prepared datasets and tuning runs are separate from this reported scope.
Uran Lajci · ★ 0 · CC0-1.0 · Jev, Clef
Author-run records; no fresh model inference reproduced
Source: Licence → Source: Zero-shot report → Source: Saved score table → Source: Jev English predictions → Source: Inference code →
Source: WLojek/nopescope ↗A Chrome extension with local Basal topic judgments, word and author rules, and MobileCLIP image matching. Its report records browser-model parity and end-to-end fixture and live-page checks on an M4 Pro. Text evaluations use synthetic English and Polish posts; image scene matching does not reliably identify people. Downloaded weights keep their own licences.
WLojek · ★ 0 · Apache-2.0 · Basal
Author-run records; no fresh model inference reproduced
Source: Licence → Source: Run report → Source: Browser inference →
Source: logesh4v/kiro-crew-decider-sagemaker ↗Deploys Strands Decider 2B hobson-v19 to a SageMaker real-time endpoint and connects it to Kiro Crew. The pinned manifest matches hobson-v19 provenance, including the adapter and head hashes. The maker reports 35 benchmark invocations with no errors on ml.m5.2xlarge in Mumbai: warm median 2,894 ms and p99 3,031 ms. These deployment measurements belong to hobson-v19, separate from the v1 three-run weight average. The default 1,000 ms client timeout needs raising. The endpoint was removed after the run. Source includes the serving shim, invocation check, benchmark and cleanup; this does not establish Bedrock availability.
S Logesh · ★ 0 · MIT · Strands Decider 2B hobson-v19
Source: Licence → Source: Serving shim → Source: Pinned weight manifest → Source: hobson-v19 checkpoint provenance →
Source: maths-oliveira/blabla-meter ↗A Chromium extension that scores pages and sections with Jev, while a generative model writes summaries on demand. Its usefulnews component publishes a 24-article Jev/LLM comparison and preference-feedback examples from September; those records do not benchmark the current extension.
maths-oliveira · ★ 0 · MIT · Jev
Maker's recorded runs · VENDOR-REPORTED; not independently reproduced
Source: Recorded Jev run → Source: Licence →
Source: nwang783/phishbench ↗Tests OpenAI Decisions API on reduced web-page text, with plain-HTML and headless-browser views. The published rendered-page summary records 60/68 reviewed phishing pages detected at threshold 0.5 and 2/198 benign pages flagged. Review labels came from an AI assistant, and raw pages and per-site predictions are not public.
Nathan Wang · ★ 0 · MIT · Decisions API
Maker's recorded runs · VENDOR-REPORTED; not independently reproduced
Source: Saved aggregate results → Source: Licence →
Source: nicolasambin/lazycontroller ↗An OpenAI-client wrapper that sends closed structured questions to a decision model and falls back on low confidence or unsupported requests. Its saved notebook shows Jev/GPT routing and a local Kev/Ollama path. Shadow mode measures agreement with the slower model; that agreement is not ground-truth accuracy.
Nicola Sambin · ★ 0 · Apache-2.0 · Jev / Kev
Maker's recorded runs · VENDOR-REPORTED; not independently reproduced
Source: Executed notebook → Source: Licence →
Source: ferronicardoso/mcp-laya ↗Local CPU inference through laya-ts and ONNX, exposed as MCP tools for typed text questions. The maker records a thank-you note changing urgency when options are reordered. Optional stability checks re-ask the question and flag flips; their scores are uncalibrated, and stable answers can still be wrong.
ferronicardoso · ★ 0 · MIT · Laya
Author-run records; not independently reproduced
Source: Licence → Source: Model integration → Source: Real-model test path →
Source: Khelendrameena/hybrid-shorts-generator ↗Selects transcript windows with text-only Laya-multilingual and deterministic features, then asks Ollama for titles and metadata. The maker records real Hindi/English Laya calls and a synthetic end-to-end render. Hook questions disagree on one clip. Live target-machine Whisper GPU and Ollama stages remain untested in that record.
Khelendrameena · ★ 0 · MIT · Laya
Author-run records; not independently reproduced
Source: Licence → Source: Recorded execution and limits → Source: Laya integration →
Source: epifaneia/callscore ↗Reviews call transcripts through closed-option Jev questions, with code measuring interruptions, copying evidence turns and calculating quality scores. The saved SQLite database records 120 artificial calls through OpenRouter. A simulated engine is the default; uncertain critical judgments go to review. These calls do not establish performance on real customers.
epifaneia · ★ 0 · Apache-2.0 · Jev
Author-run records; not independently reproduced
Source: Licence → Source: Jev adapter → Source: Saved call database → Source: Published report →
Source: lightshadow1/policy-pal-guardrails ↗Compares frozen regex rules, Jev and GLiNER on four labels. The published reports record model calls and threshold tuning on a random split of a mixed-source question pool. The two “weeks” are not later production traffic; the higher round-2b result is post hoc. The score cache is not committed.
lightshadow1 · ★ 0 · MIT · Jev
Author-run records; not independently reproduced
Source: Licence → Source: Run report → Source: Threshold and hybrid limits → Source: Model client →
Source: GrntKim/FastABC ↗Jev checks individual rubric cells, Gemini reviews uncertain cells, and graders confirm scores. Published raw calls cover 24 synthetic answers and 150 labelled cells. Ten cells go to human review; evidence lookup is billed separately. The small comparison does not establish an accuracy gain.
GrntKim · ★ 0 · MIT · Jev
Author-run records; not independently reproduced
Source: Licence → Source: Recorded evaluation → Source: Jev adapter →
Source: subnetMusk/probeloop ↗Uses a fine-tuned Laya ONNX classifier to find explicit knowledge or experience limits in Italian practice answers. A recorded synthetic HTTP/WebSocket turn uses the actual classifier and scripted LLMs. Interview decisions and evaluation remain separate; real-speech quality and physical Mac performance are still open.
subnetMusk · ★ 0 · Apache-2.0 · ProbeLoop Laya stated limits
Author-run records; not independently reproduced
Source: Licence → Source: Pilot and integration run → Source: Current acceptance limits →
Source: DarvinX/d1-pi4-asm ↗ARM64 assembly inference for d1-2B Q4_K_M and its Q8_0 projector on a 4 GB Pi 4B. Source, manifests and 80 answers are public. Choice questions update 48 selected image positions; direct counts keep all 300. Same-input timing uses four Pi pairs per format; the complete reference accuracy run uses x86 llama.cpp.
DarvinX · ★ 0 · MIT code; Liquid model and CV-Bench terms separate · d1
Author-run records; not independently reproduced
Source: Licence → Source: Benchmark protocol and results → Source: Third-party terms → Source: Image-question helper →
Source: AIAnytime/snapcall ↗A CLI and dashboard for preparing labelled examples, training decision heads, evaluating predictions and serving a System One endpoint. Its released Laya medical-document classifier includes a separate held-out test report. The toolkit’s RLCD recipe is its own implementation; it does not disclose TypeSafe’s training code.
AI Anytime · ★ 0 · Apache-2.0; model and dataset terms separate · Laya
Public trained weights and evaluation record; no fresh training run
Source: Licence → Source: Released model and report → Source: Held-out evaluation →