jev-ai.pro

mx

route3.mx.cloudflare.net

route1.mx.cloudflare.net

route2.mx.cloudflare.net

ns

gannon.ns.cloudflare.com

zara.ns.cloudflare.com

caa

txt

google-site-verification=qGmJEBYMriEqCMDHaMtudeiqlAswbR2ZZzeDqRpeLEk

v=spf1 include:_spf.mx.cloudflare.net include:spf.privateemail.com ~all

dmarc

v=DMARC1; p=reject; sp=reject; pct=100; adkim=r; aspf=r;

User-agent: *
Allow: /
Disallow: /api/
Disallow: /auth/
Disallow: /account
Disallow: /admin

Sitemap: https://jev-ai.pro/sitemap.xml

# Jev AI

> Use typed questions to evaluate text and return probabilities, choices and scores.

## Get started
- [Playground](https://jev-ai.pro/)
- [Jev AI Playground guide](https://jev-ai.pro/playground)
- [Documentation](https://jev-ai.pro/docs)
- [Use cases](https://jev-ai.pro/use-cases)
- [API documentation](https://jev-ai.pro/jev-api)
- [jevgrep: install the CLI and agent skill](https://jev-ai.pro/jevgrep)
- [Batch processing](https://jev-ai.pro/batch)
- [Pricing](https://jev-ai.pro/pricing)
- [What is Jev?](https://jev-ai.pro/what-is-jev)

## Guides
- [How to get a Jev API key](https://jev-ai.pro/jev-api-key): create a key, store it as JEV_AI_API_KEY and verify it.
- [Is Jev free?](https://jev-ai.pro/jev-free): welcome credits, credit cost per model and when you pay.
- [Is Jev multimodal?](https://jev-ai.pro/jev-multimodal): which models read images, audio and video.
- [GPT-6 Luna](https://jev-ai.pro/model/gpt-6-luna): OpenAI's Decisions API model with Jev's request format, text and images.
- [Can you run Jev locally?](https://jev-ai.pro/run-jev-locally): open-source status and self-hosted alternatives.
- [Jev vs LLM](https://jev-ai.pro/compare/jev-vs-llm): how Jev differs from chat LLMs and BERT classifiers.
- [Jev vs Claude](https://jev-ai.pro/compare/jev-vs-claude): Banking77, routing and phishing measurements against Opus and Haiku.
- [Jev vs ChatGPT](https://jev-ai.pro/compare/jev-vs-chatgpt): price, latency and TypeSafe’s eval against GPT-6 Astra and GPT-5.6 Terra.
- [Jev vs BERT](https://jev-ai.pro/compare/jev-vs-bert): zero-shot decisions against fine-tuned encoder classifiers.
- [Jev-as-a-Judge](https://jev-ai.pro/jev-as-a-judge): the measured evidence, our own 300-verdict HaluEval test, and how to set up a typed judge with thresholds.
- [Jev Search](https://jev-ai.pro/jev-search): how the open-source Jev Search picks sources and ranks results, the published retrieval numbers, and our own BM25 reranking test on SciFact.
- [Jev 1.13.0 model guide](https://jev-ai.pro/model/jev-1.13): version, context window, rate limits, pricing and JevBench standing.
- [Jev for coding agents](https://jev-ai.pro/jev-for-coding-agents): Claude Code, Codex and OpenCode setup.
- [Jev agent](https://jev-ai.pro/jev-agent): where Jev sits in an agent loop, how to call it from HTTP, the SDK, Vercel AI SDK and LangChain, public cases, and our 440-decision tool-routing test.
- [Jev examples](https://jev-ai.pro/jev-examples): 15 open-source projects built on Jev, from Pokémon Red to code linting, and our own jevlint run.
- [Jev on Ollama](https://jev-ai.pro/jev-ollama): Ollama's /v1/systemone endpoint and its decision models, measured against hosted Jev on judging and tool routing.
- [Jev in OpenClaw](https://jev-ai.pro/jev-openclaw): the decisionModel role and the official TypeSafe plugin, setup, limits and our plugin test.
- [Jev in n8n](https://jev-ai.pro/jev-n8n): the community nodes compared, Base URL settings for Jev AI keys, and a tested 40-ticket workflow.
- [Where to run Jev](https://jev-ai.pro/where-to-run-jev): TypeSafe, OpenRouter, Vercel AI Gateway, Cloudflare Workers AI, Bedrock, Vertex and Azure.
- [Jev 模型中文指南](https://jev-ai.pro/zh/jev-model)

## Other languages
- [Jev API 中文文档](https://jev-ai.pro/zh/jev-api) · [Jev AI 价格](https://jev-ai.pro/zh/pricing)
- [API de Jev en español](https://jev-ai.pro/es/jev-api) · [Precios de Jev AI](https://jev-ai.pro/es/pricing)
- [API do Jev em português](https://jev-ai.pro/pt/jev-api) · [Preços do Jev AI](https://jev-ai.pro/pt/pricing)

## Model guides
- [Kev](https://jev-ai.pro/model/kev): the open Jev alternative you can run and fine-tune; Kev-4B on a laptop against hosted Jev on 300 HaluEval verdicts.
- [pplx-decider](https://jev-ai.pro/model/pplx-decider): Perplexity’s open 27B decision model; its panel as a radar, the clinical claim examined, and our three-way HaluEval test against Jev and Kev.
- [Jeeves](https://jev-ai.pro/model/jeeves): PostHog's open reasoning decision model; its claims against Jev and our test on a Mac with and without thinking.
- [Microsoft-Decision-1](https://jev-ai.pro/model/microsoft-decision-1): Microsoft's hosted decision model; its 36-benchmark claims and our same-day test against Jev on 300 HaluEval verdicts and 440 tool-routing decisions.
- [Clef-Omni](https://jev-ai.pro/model/clef-omni): Cloudflare's multimodal decision model, in the Jev AI playground for images and WAV or MP3 audio; photos and spoken digits against Jev AI's Jev-Omni, and text decisions against Jev.
- [Imajev-4B](https://jev-ai.pro/model/imajev): Jev’s request format, open weights, and two photos per request.
- [JevK5](https://jev-ai.pro/model/jevk5): The fastest open 4B on the board, with the weights published.
- [Winnow-12B](https://jev-ai.pro/model/winnow-12b): One local model that decides, chats and looks at a screenshot.
- [djev](https://jev-ai.pro/model/djev): Decisions read off a diffusion model in one denoising step.
- [decider-4b](https://jev-ai.pro/model/decider-4b): Millisecond decisions from 8.4 GB of open weights, on TypeSafe’s wire format.
- [Cygnet](https://jev-ai.pro/model/cygnet): Seventh on the board, and nothing about the model is trained.
- [SemIf](https://jev-ai.pro/model/semif): Semantic ifs from a frozen open model, with no generation at all.
- [OpenJev](https://jev-ai.pro/model/openjev): One name, ten projects. Find yours, then run the drop-in.

## Evaluation & quality
- [Browse category](https://jev-ai.pro/use-cases/evaluation)
- [LLM as a Judge](https://jev-ai.pro/llm-as-a-judge): Evaluate an answer for source support, relevance and quality using your own rubric.
- [AI Agent Evaluation](https://jev-ai.pro/ai-agent-evaluation): Check completion claims against tool results, task requirements and allowed actions.
- [Claude Code Rules Checker](https://jev-ai.pro/claude-code-rules-checker): Check a code diff against each project rule and get complies, violates or insufficient evidence per rule.

## Data matching
- [Browse category](https://jev-ai.pro/use-cases/data-matching)
- [Entity Matching](https://jev-ai.pro/entity-matching): Compare product or organization records with a same, different or review decision.

## Retrieval & knowledge
- [Browse category](https://jev-ai.pro/use-cases/retrieval)
- [RAG Evaluation](https://jev-ai.pro/rag-evaluation): Check each retrieved passage for relevance, answer coverage and injected instructions before generation.
- [Live Web Context](https://jev-ai.pro/web-context): Search the web for a yes/no question and see how live evidence changes the answer.

## Routing & decisions
- [Browse category](https://jev-ai.pro/use-cases/routing)
- [LLM Router](https://jev-ai.pro/llm-router): Choose a model tier, handler or tool for each request, with confidence to fall back safely.
- [MCP Tool Router](https://jev-ai.pro/mcp-tool-router): Pick the next MCP tool for an agent task, including no tool at all and actions that need human confirmation.
- [Support Ticket Triage](https://jev-ai.pro/support-ticket-triage): Classify a support ticket, assign a team and set its priority with your own labels.

## Safety & security
- [Browse category](https://jev-ai.pro/use-cases/safety)
- [Prompt Injection Detector](https://jev-ai.pro/prompt-injection-detector): Check text for instruction overrides, prompt injection and data-exfiltration attempts, with a review recommendation.

## Classification & scoring
- [Browse category](https://jev-ai.pro/use-cases/classification)
- [Lead Qualification](https://jev-ai.pro/lead-qualification): Compare a sales lead with your ideal customer profile to judge fit, buying intent and the next follow-up.
- [Sales Call Scoring](https://jev-ai.pro/sales-call-scoring): Score a call transcript against your own scorecard, item by item, and see which topics were never covered.
- [Jev Jobs](https://jev-ai.pro/jevjobs): Jev job search: import a resume, find jobs posted in the last 30 days and score each one on your weighted criteria.

## Policies
- [Privacy](https://jev-ai.pro/privacy)
- [Terms](https://jev-ai.pro/terms)