Skip to content
cccskills
FreePublished on GitHub

model-benchmarker

Use when: benchmark model latency, throughput, or quality.

The English summary is being prepared. Showing the original description.

See how to install

Files included (1)

  • SKILL.md1.8 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Model Benchmarker

Tests Model-Performance über mehrere Provider (OpenRouter, lokale Ollama-Instanzen).

Tests

TestBeschreibungRunsMetrik
LatenzZeit bis erster Token (kleiner Prompt)10Median in Sekunden
DurchsatzTokens pro Sekunde (4K Prompt)5Median tok/s
QualitätAntwort auf 4 Testfragen bewertet1Score 0-100%

CLI-Verwendung

# Einmaliger Test
python model-benchmarker.py --run openrouter/deepseek-v4-flash
python model-benchmarker.py --run "local/qwen3.5:9b"

# Alle Provider testen (Default-Modell pro Provider)
python model-benchmarker.py --all

# Vergleichstabelle aller getesteten Modelle
python model-benchmarker.py --compare
python model-benchmarker.py --compare --json   # JSON-Ausgabe

# Trend-Historie
python model-benchmarker.py --history openrouter
python model-benchmarker.py --history local

Provider-Konfiguration

Liest automatisch aus config.yaml:

  • model.default → OpenRouter-Modell
  • custom_providers → Lokale Ollama-Instanzen und deren Modelle
  • API-Keys aus .env (OPENROUTER_API_KEY)

Ergebnisverzeichnis

Alle Ergebnisse in ~/.model-benchmarks/results/<provider>-<model>.json mit vollständiger Historie (max 50 Einträge).

Cron-Job

Wöchentlicher --all Run ist als Cron-Job eingerichtet:

  • Führt alle Provider-Modelle durch
  • Sendet Ergebnisse an Terminal (lokal)
  • Kein Agent nötig (no_agent: true)

Abhängigkeiten

Keine — nur Python-Standardbibliothek (urllib, json, statistics, etc.)

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Build integrated IS/BS/CF financial workbooks in Excel.

The English summary is being prepared. Showing the original description.

openamer/openamer8Updated Oct 20, 2026

Run and grow the OpenAmer Agent-to-Agent swarm: identity, trust, node-to-node ask, signed skill/insight sharing, and the autonomous self-learning loop.

The English summary is being prepared. Showing the original description.

openamer/openamer8Updated Oct 20, 2026

Use for A/B experiments on OpenAmer configs and skills.

The English summary is being prepared. Showing the original description.

openamer/openamer8Updated Oct 20, 2026

Roleplay a hostile user to find and triage UX pain points.

The English summary is being prepared. Showing the original description.

openamer/openamer8Updated Oct 20, 2026

Agent mesh: master/worker nodes, HTTP delegation, heartbeat.

The English summary is being prepared. Showing the original description.

openamer/openamer8Updated Oct 20, 2026

Use for agentic commerce payments tasks. Candidate synthesized from trend radar; awaiting trial promotion.

The English summary is being prepared. Showing the original description.

openamer/openamer8Updated Oct 20, 2026

See all skills by openamer

Report a problem with this skill