Will it save me money on day one in fixture mode?

No. It starts safely in fixture mode, makes no paid calls and makes no savings claim; synthetic profiles test behavior only, and real routing needs measured profiles, calibrated checks, representative labels, and a promotion decision.

What does contract_passed actually guarantee?

Only that configured gates passed, not guaranteed factual truth. The system permits at most two generation attempts and verifies the final attempt as well.

What happens if a provider times out mid-billing?

A network timeout preserves unknown cost instead of treating it as zero, with atomic reservations and durable audit leases in SQLite WAL.

What stops smoke_providers.py from spending real money?

An explicit budget. A real smoke is opt-in and refuses to start without both credentials and a cap: python scripts/smoke_providers.py --real --max-budget-microusd 10000. Copy .env.example and keep credentials server-side, and never set mode active with fixture profiles, since real routing needs measured profiles, calibrated checks, representative labels, and a promotion decision.

What happens if the first generation attempt fails?

It permits at most two generation attempts and verifies the final attempt as well, so a retry does not skip the checks; the response records the attempt count alongside the checks and cost status.

How do I run the API, audit worker and dashboard?

uvicorn app.api:app serves the API, python -m app.worker runs the audit worker, streamlit run dashboard/app.py opens the dashboard, and docker compose up starts the containers with --profile dashboard adding Streamlit.

What do I need before switching to active routing?

Measured profiles, calibrated checks, representative labels, and a promotion decision; the README warns against setting mode active with fixture profiles.

CommunityJev Tools & CLIs

LLM Cost Optimizer Jev

Fixture-first LLM router that picks the lowest expected-cost qualified route, verifies output before release, and records cost provenance with idempotency keys.

See highlights

Open github.com

Stars: 0
Language: Python
License: APACHE-2.0
Published by: Community
Author: lorensation
Last verified: Oct 2026

lorensation/llm-cost-optimizer-jev

User repository on GitHub

View on GitHub

An intelligent routing layer powered by TypeSafe AI's System One model Jev that sits in front of multiple LLM providers, analyzes each incoming request’s complexity, routes it to the cheapest model capable of handling it at acceptable quality, and continuously validates that routing decisions are correct.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

An evidence-based LLM router that chooses the lowest expected-cost qualified route, verifies every output before release, and records complete cost provenance. The repository starts safely in fixture mode: it makes no paid calls and makes no savings claim. The response includes the immutable policy snapshot, model, checks, attempt count, and cost status. contractpassed means all configured gates passed; it is not a claim of guaranteed factual truth.

Quoted from the project's README. Read the full README.

Tagscommunity github

Works forJev for Routing

Used inJev for Software & DevTools

Languages: Python

Highlights

Starts in fixture mode with no paid calls; real smoke needs credentials plus an explicit microdollar budget flag.
Response carries policy record, model, checks, attempt count, and cost status; contract_passed means gates passed.
SQLite uses WAL with atomic reservations, durable audit leases, and idempotency keys; at most two attempts.
Serves API via uvicorn, worker via python -m app.worker, dashboard via Streamlit, or docker compose.

Watch out

Experimental: synthetic profiles and fixture outputs test behavior only; real routing needs measured profiles and labels. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

pyGPT

A desktop AI assistant with a Jev / System One inline plugin for classification, routing, verification, and scoring without chat-style output.

Jev Tools & CLIs#community#python#desktop

Community

szczyglis-dev/py-gptPython

horosa-skill

452

An AGPL MCP server and JSON-first CLI exposing 110 astrology and metaphysics techniques from Horosa to local AI clients, running offline on macOS and Windows with SQLite-backed call history.

Jev Tools & CLIs#community#jevbest

Community

Horace-Maxwell/horosa-skillPython

System One Harness

209

A Python harness for System One models. It runs Jev and other decision models locally, or through HarnessRouter.ai.

Jev Tools & CLIs#community#python#open-models

Community

HarnessRouter/SystemOneHarnessPython

Harness Router

MIT-licensed Python tool router for coding agents that keeps obvious calls on a fast path, asks Jev (default typesafe/jev-1.13 via OpenRouter) for genuinely ambiguous choices, and can run bounded MCTS when multi-step consequences matter. Ships a CLI, MCP server and hook integrations.

Jev Tools & CLIs#community#routing#mcp

Community

Protocol-Lattice/harness-routerPython

jevclip

Video tool where Jev scores subtitle segments and code assembles a short highlights cut plus a de-watered full cut with timestamped summaries.

Jev Tools & CLIs#community#jevbest#catalog-import

Community

cclank/jevclipPython

jev-yaba-wechat

macOS floating WeChat companion that reads the visible chat with on-device Vision OCR, judges intent and reply risk with Jev, drafts toned replies with GPT, and only fills text after you approve it.

Jev Tools & CLIs#community#jevbest#catalog-import

Community

wuxie888/jev-yaba-wechatPython

From the Jev guides

Original write-ups on what this entry is for.

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.

Jev Tools & CLIs#gateway#vercel#cloudflare

Read article

Exporting Decision Models: GGUF, ONNX, and Device Runtimes

Oct 10, 2026

Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.

Jev Tools & CLIs#open-models#local#onnx

Read article

Records at Rest: Documents, Inboxes, and Risk Queues

Sep 28, 2026

Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.

Jev Tools & CLIs#classification#email#security

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai