Local Model Spend Router v2

Released: 2026-04-29

Price: $4.99

Category: analytics

What changed from v1

Not a duplicate: v1 is an interactive single-task router driven by CLI flags such as --task, --complexity, --privacy, and remaining budget. v2 is a batch/config-as-code edition: it reads task backlogs and a model catalogue from JSON, chooses the cheapest suitable model for each task, and reports total paid spend before any model is called.

What it does

local-model-spend-router v2 takes the guesswork out of AI model spend at scale. Define your task backlog as a JSON file, supply your model catalogue with real cost and quality data, and let the router assign the cheapest model that meets each task's requirements — local models for low-risk work, paid tiers only when quality or privacy demands it.

Unlike v1 which routes one task at a time via CLI flags, v2 processes entire backlogs in one run and shows you the total paid spend estimate before any model is called. The config-as-code approach means routing policies live in version control alongside your workflows.

Use cases

• Process overnight task backlogs through the cheapest suitable models automatically.
• Maintain a custom model catalogue with your real cost and quality ratings.
• Get a total paid spend estimate before committing any API budget.
• Version-control your routing policies alongside your automation code.
• Run batch routing for CI/CD pipelines, nightly reports, and bulk analysis jobs.

Requirements

• OpenClaw v2026.3.23 or later recommended.
• Python 3.10+.
• No third-party Python packages required.
• No API keys required — works with your existing model setup.

Example usage

python3 scripts/route_models.py \
  --tasks tasks.json \
  --models models.json \
  --output routing-decisions.json

tasks.json example

[
  {"id": "t1", "description": "Summarise overnight logs", "complexity": "low", "privacy": "internal"},
  {"id": "t2", "description": "Generate compliance report", "complexity": "high", "privacy": "sensitive"},
  {"id": "t3", "description": "Classify support tickets", "complexity": "medium", "privacy": "internal"}
]

QA and security notes

Viper QA passed 6/6 and Warlock security cleared this release on 2026-04-29.