GemmaMatch — Gemma 4 Local Hardware Matcher

Auto-detect hardware and recommend the best Gemma 4 model for local deployment on PC, Mac, or mobile.

Security Scan

The skill is internally consistent: it is an instruction-only browser-based hardware detector that asks users to visit a website to run local WebGPU/WebGL detection and then provides model recommendations and copy-paste run commands.

Purpose & Capability

The name and description (auto-detect hardware and recommend Gemma 4 variants) align with the SKILL.md content. The skill is instruction-only and uses browser WebGPU/WebGL detection and platform-specific run commands — these are coherent with the stated purpose.

Instruction Scope

SKILL.md instructs users to visit the website and allow in-browser hardware detection; it does not instruct the agent to read system files, env vars, or send data elsewhere. Note: detection happens in the user's browser (client-side) rather than the agent, so the skill relies on the website and user consent to run detection.

Install Mechanism

No install spec or code files are present (instruction-only), so nothing is written to disk or downloaded by the skill itself.

Credentials

The skill requests no environment variables, credentials, or config paths — this is proportionate to a read-only hardware-detection/recommendation guide.

Persistence & Privilege

The skill is not always-on and uses default model invocation settings; it does not request elevated or persistent system privileges.

Assessment

This skill appears coherent, but exercise normal caution before acting on any recommended terminal commands: 1) Confirm the website uses HTTPS and inspect the GitHub source (https://github.com/walex8925/Gemma4local) if you can. 2) Review any copy-paste run commands before executing — avoid piping unknown scripts into a shell (e.g., curl | sh). 3) Verify the site’s privacy claim by checking whether detection scripts make network calls (you can inspect the page or the repo). 4) If you want complete assurance, open the repository code locally or in a sandboxed browser to validate that hardware detection runs purely in-browser and no external data is exfiltrated.

Find the best Gemma 4 model for your hardware in seconds.

Website: https://www.gemmamatch.com

What it does

GemmaMatch auto-detects your GPU, VRAM, and system specs via WebGPU/WebGL APIs, then recommends the most suitable Gemma 4 model tier and provides a ready-to-use run command. All processing happens locally in your browser — no data leaves your device.

Recommended model tiers

Tier	Target hardware	Use case
Gemma 4 E2B	Phones, tablets, low-VRAM devices	On-device inference, edge deployment
Gemma 4 26B MoE	Desktop GPUs (8-16 GB VRAM)	General local AI, coding assistance
Gemma 4 31B Dense	Workstations (24+ GB VRAM)	High-quality generation, research

Key features

• Automatic GPU detection — uses WebGPU and WebGL APIs, no install required
• Personalized model recommendation — matches your exact hardware to the optimal Gemma 4 variant
• Platform-specific setup guides — step-by-step instructions for Mac (MLX, Ollama), Windows (Ollama, LM Studio), iOS, and Android
• One-click run commands — get a copy-paste Ollama or LM Studio command tailored to your system
• Manual comparison mode — compare upgrade scenarios or override auto-detection
• Privacy-first — everything runs in-browser, zero data collection

Quick start

1. Visit https://www.gemmamatch.com
2. Allow hardware detection (or enter specs manually)
3. Get your recommended model + run command
4. Copy the command and run it in your terminal

Supported platforms

• macOS — Apple Silicon (M1-M4), Intel with discrete GPU
• Windows — NVIDIA (RTX 30/40/50 series), AMD (RX 7000 series)
• Linux — NVIDIA CUDA, AMD ROCm
• iOS / Android — on-device model recommendations

Links

• Website: https://www.gemmamatch.com
• GitHub: https://github.com/walex8925/Gemma4local
• Product Hunt: https://www.producthunt.com/products/gemma-4-local-hardware-matcher