ypollak2/llm-router

v8.5.0 Feature

This release adds 2 notable features for engineering teams evaluating rollout.

Published 20d LLM Frameworks

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai-routing anthropic claude claude-code cost-optimization gemini

+7 more

litellm llm llm-router mcp-server model-router ollama openai

ReleasePort's take

Light signal

editorial:auto 13d

v8.5.0 ships Savings Dashboard with token routing metrics and time-series views. Simultaneously corrects calculation errors: Opus-to-Sonnet 4.6 baseline shift, flat-rate quota display, and classifier overhead attribution.

Why it matters: Cost-tracking users should review the dashboard to verify accurate savings post-corrections. Prior calculations were misleading; this update enables honest cost optimization analysis across 14d-to-all-time windows.

Summary

AI summary

Savings Dashboard adds token-centric analytics with ANSI colors and time‑series breakdowns.

Changes in this release

Type	Severity	Summary	CVE
Feature
Feature	Medium	New Savings Dashboard displays token routing metrics with ANSI visualization New Savings Dashboard displays token routing metrics with ANSI visualization Source: llm_adapter@2026-05-21 Confidence: high	—
Feature	Medium	Dashboard supports 14d, 30d, 3m, 1y, and all-time windows Dashboard supports 14d, 30d, 3m, 1y, and all-time windows Source: llm_adapter@2026-05-21 Confidence: low	—
Feature	Medium	Dashboard supports 14‑day, 30‑day, 3‑month, 1‑year, and all‑time windows Dashboard supports 14‑day, 30‑day, 3‑month, 1‑year, and all‑time windows Source: granite4.1:30b@2026-05-22-audit Confidence: low	—
Bugfix
Bugfix	Medium	Default baseline changed to Sonnet 4.6 from Opus for honest comparison Default baseline changed to Sonnet 4.6 from Opus for honest comparison Source: llm_adapter@2026-05-21 Confidence: high	—
Bugfix	Medium	Flat-rate Claude users see quota freed, not inflated dollar savings Flat-rate Claude users see quota freed, not inflated dollar savings Source: llm_adapter@2026-05-21 Confidence: high	—
Bugfix	Medium	Classifier overhead subtracted from net savings calculation Classifier overhead subtracted from net savings calculation Source: llm_adapter@2026-05-21 Confidence: high	—

Full changelog

What's New

Savings Dashboard (`llm_dashboard`)

Token-centric dashboard with ANSI colors, time-series, and honest metrics:

╔══════════════════════════════════════════════════════════════╗
║  LLM Router — Savings Dashboard                              ║
╠══════════════════════════════════════════════════════════════╣

  LAST 14 DAYS

  142.5K tokens routed  ·  95 calls  ·  $2.34 net saved

  Tokens saved per day:
  ▁▃▅▂▆▇█▅▃▄▆▅▇▆

  BREAKDOWN         Calls   Tokens      Gross $     Net $
  ─────────────────────────────────────────────────────
  ollama (FREE)       47    71.2K      $1.41      $1.41
  gemini              23    34.1K      $0.69      $0.67
  deepseek            12    22.0K      $0.31      $0.30
  Classifier overhead:                            -$0.03
  NET SAVED:                                      $2.34

  ████████████████████ ollama     49%
  █████████████       gemini     24%
  ████████            deepseek   13%

Audit Corrections

Honest baselines: Sonnet 4.6 as default (not Opus strawman). Opus still available via --baseline opus
Subscription-aware: Flat-rate Claude users see "quota freed" instead of fake dollar savings
Classifier overhead: Subtracted from net savings — no more inflated numbers
Time windows: 14d, 30d, 3m, 1y, all

Engineering Audit

Full diagnostic report at docs/AUDIT_2026Q2.md covering activation reliability, routing accuracy, metrics honesty, and visibility gaps.

Upgrade

pip install --upgrade llm-routing

Full Changelog: https://github.com/ypollak2/llm-router/compare/v8.4.0...v8.5.0

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track ypollak2/llm-router

Get notified when new releases ship.

About ypollak2/llm-router

Subscription-aware LLM router for Claude Code. Routes tasks to 20+ providers (OpenAI, Gemini, Groq, Ollama, Codex) based on complexity classification, Claude subscription pressure, and cost. Free tasks stay on Claude subscription; expensive tasks fall back to the cheapest capable model. Includes 30 MCP tools, 6 auto-routing hooks, semantic dedup cache, prompt caching, daily spend cap, and a live web dashboard.

All releases →

Related context

Related tools

Earlier breaking changes

v9.2.0 Changes auto‑route directive from advisory "DO NOT SKIP" to hard constraint with explicit blocked tools list.
v9.2.0 Breaks permanent downgrade of enforcement after first Edit/Write; v13 now requires per‑turn routing.