LocalPilot Logo LocalPilot v0.1.1
GitHub
🦙 Model Catalog & Sizing

Find the AI That Fits Your Computer

Different models have different strengths and memory requirements. Here is how to choose the right one for your machine.

What Does "3B", "7B", or "8B" Mean?

The letter "B" stands for Billion Parameters. In simple terms, parameters are the connections in the AI's virtual neural network:

Smaller Models (1B – 3B)

Lightning fast, use very little RAM (2GB to 3GB), ideal for quick summaries, editing text, and older or thinner laptops.

Standard Models (7B – 9B)

The "sweet spot" for reasoning, complex instructions, and creative writing. Requires about 5GB to 8GB of free memory.

Tier 1

Everyday Laptops (4 GB – 8 GB RAM)

Meta ~2.0 GB RAM

Llama 3.2 (3B)

Meta's flagship compact model. Incredibly fast, great at summarization, proofreading, and natural dialogue.

localpilot chat llama3.2:3b
Alibaba ~2.1 GB RAM

Qwen 2.5 (3B)

Extremely smart instruction-following and strong multilingual capabilities. Perfect for non-English inquiries.

localpilot chat qwen2.5:3b
Microsoft ~2.3 GB RAM

Phi-3 Mini (3.8B)

Microsoft's reasoning powerhouse. Trained on curated textbook-quality data for logic and academic questions.

localpilot chat phi3:mini
Tier 2

Balanced Workstations (8 GB – 16 GB RAM)

Meta ~4.7 GB RAM

Llama 3.1 (8B)

The gold standard for general intelligence. Excellent storytelling, research analysis, and creative ideation.

localpilot chat llama3.1:8b
Mistral AI ~4.1 GB RAM

Mistral (7B)

Renowned European model. Crisp, concise responses, excellent technical explanations and structured formatting.

localpilot chat mistral:7b
Google ~5.4 GB RAM

Gemma 2 (9B)

Google's high-efficiency open model. Outstanding benchmarks for factual knowledge, reading comprehension, and math.

localpilot chat gemma2:9b
🛡️

The LocalPilot Safety Shield

You never have to calculate memory math manually. Whenever you run localpilot recommend, our hardware scanner inspects your available RAM and marks models with a safe compatibility percentage so your system never runs out of memory.