Agents: read the raw Markdown of this page, or start at llms.txt.
~/wiki/ideas
Open replicas: servers and logit readers over other models
TL;DR None of these is TypeSafe or runs Jev. Each wraps a model it did not train (a stock LLM read by option-letter logits or logprobs, an NLI or embedding model, a diffusion LM, or an LLM asked to write probabilities) behind Jev's request shape or something close to it, and each computes
confidenceits own way, so thresholds tuned on Jev do not transfer. Several answer tojev-latest: check the responsemodel. Read the Warnings on Warnings: not-Jev services, key safety, look-alikes and install names first (key routing, look-alikes, install names); models with their own weights, the Laya family, the which-is-which table and the replicas' own negatives are there too. Split from Open replicas and Jev-compatible servers on 2026-09-25 (nothing dropped).
Numbers are each author's unless marked. ★ · push as captured 2026-09-23 to 25. Pattern IDs: Decision patterns from the community (with fit verdicts). "Our check" = our 2026-09-25 source and registry vetting (docs/sweep record), not a raw capture. JevBench ranks are by version (Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites).
Servers and readers
| Tool (owner) | What | Install | ★ · push | Their numbers and caveats |
|---|---|---|---|---|
| jeff (Logan Markewich) | /v1/systemone over GLiFormer (knowledgator, 400M); accepts jev-latest; 529 on a full queue, 429 with retry-after-ms |
clone | 246 · 2026-09-20 · MIT | Jev run live: Jev far ahead on accuracy, jeff cheaper and slower under load (Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites). JevBench v1.2.2: jeff 66.9 (#9 of 18), below Jev (board row and hard tier on Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites). v1.4.2: #40, 30.58. Self-run hard items: Jev lost the temporal and numeric items (matches Jev 1.13 jaggedness: known failure modes) and led the ambiguous ones (Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites) |
| localjev (GitHub Next) | Bun bridge: any OpenAI-compatible local model writes probabilities as JSON (self-reported, not logits); accepts jev-latest, jev-preview; 529 above 64 queued |
clone | 775 · 2026-09-18 · MIT | M5 Max bake-off, 120 items: short-input macro Qwen3.6-35B-A3B 76.7%. 2,048 unrelated words lowered every model's macro accuracy (Qwen 76.7% → 69.2%; Gemma E4B SST-5 50.0% → 12.5%). Evidence that prompted-probability replicas degrade under irrelevant state. Answers as localjev-0.2 (only the response model tells); the authors say to check calibration yourself; no Jev comparison |
| OpenJev (razorback16) | DiffusionGemma 26B-A4B via patched vLLM or MLX, read as a diffusion canvas; also serves Laya and Verdict; extensions incl. images; accepts jev-latest |
clone; hosted free on Codiv (100M input tokens) | 421 · 2026-09-24 · Apache-2.0 | Author, one GPU: p50 27 ms (1 question), 31 ms (3); 57.4 req/s at concurrency 64 (p95 1,109 ms). JevBench v1.4.2 #27 (36.85; thinking mode #68). Codiv key routing: Warnings: not-Jev services, key safety, look-alikes and install names (c) |
| NotJev (Nathanael Braun) | letters the options, asks any logprob endpoint for one token (26 options max); abstains below theta |
npm i notjev |
21 · 2026-09-22 · Apache-2.0 | Author: Qwen3-8B Q4 p50 23 ms; 27B NVFP4 on vLLM 101 ms; its one-token readout agrees 0.947 with the same model generating under a grammar (n = 1,224; majority-class null 0.693). Its "hosted Jev 419 ms" has no source in the repo. In serve mode one key is both upstream bearer and caller guard |
| jevify (Felipe Infante de Castro) | any LLM as a Jev-like endpoint: hashed YAML recipe, state sent once, logprob ladder; endpoint, embedding, NLI or rerank backends | uv tool install jevify (PyPI) |
43 · 2026-09-23 · MIT | Unmodified typesafe-sdk works against it (author). policy-hard-52: DeepSeek-V4-Flash 47/52, Gemma 4 E4B 45/52. No Jev comparison |
| fastjev (chengyongru) | SDK-first SemIf fork: Torch, vLLM, MLX, llama.cpp GGUF, EXL3; 2–16 options; rejects jev-latest |
pip install 'fastjev[torch]' (npm fastjev is unrelated) |
21 · 2026-09-24 · MIT | RTX 5090: vLLM batched 36.26 vs Torch 18.52 decisions/s. Jev side copied (SemIf's 0.845 vs Jev 0.883 on 102 rows; count once). Headline says "open source implementation" while its results say Jev is not reproduced |
| OpenDecision (Deepan Wadhwa) | zero-shot NLI (ModernBERT-large) behind the Jev shape, plus retrieval | pip install OpenDecision (Python ≥ 3.13) |
57 · 2026-09-21 · Apache-2.0 | Treat scores as uncalibrated (README). JevBench v1.4.2 #56, 21.64. P24 |
| open-alternative-jev (Iker Moel Tacher) | in-process library: all questions about one state in one pass, option-letter logits; not SDK-compatible | clone (the PyPI name in its README does not exist) | 54 · 2026-09-25 · Apache-2.0 | Stock Qwen3.6-27B 8-bit on typed-decisions: 73.7% / ECE 0.020, above Jev's card figures (copied; gold is teacher agreement: Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites). Reversing options moved 4B yes/no accuracy 13.5 pts; packing hurts below ~4B. v1.4.2 #36. P01 |
| choosekit (Felix Koba) | TS: option-label logprobs from llama.cpp, Ollama or OpenRouter, with images; choosekit-mcp |
npm i choosekit |
23 · 2026-09-24 · Apache-2.0 | Jev run live (OpenRouter) on SuperGPQA: second to Kimi K3 at a small fraction of its cost (Head-to-head benchmarks: Jev on public datasets and suites); graduate knowledge questions are a poor fit (Jev 1.13 jaggedness: known failure modes). SemIf authored set: tie with a local Qwen model (Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites) |
| OpenJevPro (Corel Zhang) | letter-logprob scoring on Ollama or OpenAI-compatible models; "guard" wrapper around Jev | pip install openjevpro |
30 · 2026-09-25 · PolyForm Noncommercial + paid licence | Author's Banking77 slices: Jev tied the open models answering directly and beat OpenJevPro's own harness; its README's Jev latency claim contradicts its own measurements, and its cost and SLA claims are unmeasured. Tiny samples; numbers on Head-to-head benchmarks: Jev on public datasets and suites |
| rizzo-flow (Simone Rizzo) | option-letter logits from one shared prefill; adds numeric and an "insufficient evidence" option |
clone | 459 · 2026-09-25 · Apache-2.0 | Author (RTX 5060 Ti): SemIf authored144 0.812 vs SemIf 0.819; p50 49 ms. 6 of 36 confident wrong answers when evidence is missing. "No Jev labels": unverified |
| djev-run (Daniel Lee) | Cloud Run deployment of mmastrac/djev on DiffusionGemma | ghcr.io image | 551 · 2026-09-24 · no licence | Author: $3.19/h on one RTX PRO 6000, scales to zero; cold start ~47.5 s; 117 ms median. Its "JevBench 73.4" is self-run, not an official row. Answer shape differs from the docs (1-indexed Score probabilities, no type field; our check) |
| SNAP (Emanuele Menon) | Rust over vendored llama.cpp; adds numeric, allow_abstain, and a confidence on Noul (Jev has none) |
build | 21 · 2026-09-25 · no licence file | Author: Qwen3.8-4B agrees 73.2% with frontier-consensus labels over 373 TypeSafe public decisions; option-order reversal flips a third of MiniCPM's choices; 51 ms vs Ollama 149 ms. Binary snap shadows Ubuntu's |
| Glance (Yohei Nakajima) | logits from a frozen Qwen3-VL-4B for image decisions; /v1/decide |
pip install glance-vlm |
23 · 2026-09-24 · Apache-2.0 | Pre-registered fresh photos: yes/no 0.939 (541) vs Gemini 3.1 Flash-Lite 0.961; exact rating on five synthetic 4-level scales (1,000 images) 0.669 vs 0.763. vs Laya Vision on an M5: 0.886 vs 0.689 (0.747 without Laya's Pets37 collapse; Laya faster) (page). P34 |
| JevEmbed (HIT Shenzhen) | any embedding model → Choice/Score/Noul by cosine and temperature; normalised-entropy confidence | clone | 26 · 2026-09-25 · Apache-2.0 | Author, JevBench public 231: best Qwen3-Embedding-8B 58.44%; that report was deleted from the default branch on 2026-09-25 (captured from commit 9cb60bb). Dataset licences mixed, incl. CC-BY-NC; distillation note: Warnings: not-Jev services, key safety, look-alikes and install names (b). P17 |
| AnyJev (Nokia Applied Research) | any HF LLM, debiased, optional head on 100–300 labels | pip install "anyjev[hf]" |
267 · 2026-09-23 · Apache-2.0 | order flips 0.230 → 0.073 (Qwen3-8B) |
| Simple Jev (Eugene Cheah) | logits server, free demo API | pip from repo | 491 · 2026-09-21 · Apache-2.0 | own contract. typed-decisions card: Qwen3.6-35B-A3B-classifier 0.716 (Jev's figure: Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites). JevBench v1.4.2: Qwen3.8-27B #32, Qwen3.6-35B-A3B #52 |
| jev-rs (Yijun Yu) | Rust over llama-server; serve, MCP, eval | README's first command is curl -fsSL …/install.sh | sh (read the script first; prebuilt binary, else a cargo build), then cargo install jev-rs with a Rust toolchain |
5 · 2026-09-22 · Apache-2.0 | SDKs reach it via TYPESAFE_BASE_URL |
| hearim (ziozzang) | Go gateway over Ollama, vLLM, SGLang | go build |
5 · 2026-09-22 · NOASSERTION | entropy-based confidence |
| Recipe (@theanandprasad) | softmax over yes/no logits | — | post 2026-09-22 | 21/22 on Llama 3.3 70B; raw logits flip with option order (AnyJev) |
Before swapping one in, test option order, irrelevant state, missing evidence and general judgement on your own labels: "What the replicas' own negatives say" on Open replicas and Jev-compatible servers and Testing and evaluating a Jev workflow.
Related
- Warnings: not-Jev services, key safety, look-alikes and install names — the Warnings block; Open replicas and Jev-compatible servers — which-is-which, models with their own weights, Laya and its servers, trust notes and the build decision
- Eval boards: JevBench by version, Jevals.com, typed-decisions and other independent suites — JevBench by version; Head-to-head benchmarks: Jev on public datasets and suites — public datasets
- Community SDKs and libraries — community SDKs and libraries that call the real Jev
- Confidence vs probability — Jev's own
confidence, which these servers do not reproduce; TYPESAFE_* environment variables across SDKs —TYPESAFE_BASE_URL
Sources
Links inline; raw captures (2026-09-23 to 25) in frontmatter. "Our check" items come from the 2026-09-25 vetting record (docs/sweep/2026-09-25-mrjev), not raw captures.