Module 1 • Lesson 3

The Model Landscape: GPT, Gemini, Claude & Open Source

Navigate the rapidly evolving landscape of foundation models and understand when to use which.

The Model Landscape

The AI model ecosystem is evolving rapidly. As a prompt engineer, understanding the strengths and trade-offs of different models is critical.

Proprietary Models: GPT-4/4o (OpenAI), Gemini 1.5/2.0 (Google), Claude 3.5/4 (Anthropic) — state-of-the-art, API-based, pay-per-token

The AI model ecosystem is evolving rapidly. As a prompt engineer, understanding the strengths and trade-offs of different models is critical.

Proprietary Models: GPT-4/4o (OpenAI), Gemini 1.5/2.0 (Google), Claude 3.5/4 (Anthropic) — state-of-the-art, API-based, pay-per-token

Open Source Models: Llama 3 (Meta), Mistral, Qwen, Gemma — self-hostable, customizable, no per-token cost

Model Families: Each provider offers tiers — flagship (expensive, powerful) vs. lightweight (cheap, fast)

Multimodal Models: Models that process text, images, audio, and video (GPT-4o, Gemini, Claude 3.5)

Model Selection Matrix: Task complexity × latency requirements × budget × data sensitivity

Benchmarking: MMLU, HumanEval, GSM8K, GPQA — standardized tests to compare models

Model	Strengths	Best For
GPT-4o	Versatile, strong reasoning	General-purpose, complex tasks
Claude 3.5 Sonnet	Long context, safety, coding	Document analysis, careful reasoning
Gemini 2.0 Flash	Fast, multimodal, cost-effective	High-volume, speed-critical tasks
Llama 3 70B	Open source, self-hostable	Privacy-sensitive, custom fine-tuning
Mistral Large	European, strong multilingual	EU compliance, multilingual tasks

Defaulting to the most expensive model for every task

Ignoring open-source options for cost-sensitive workloads

Not benchmarking on YOUR specific use case

Assuming model rankings are static — they change quarterly

Read to the end of the lesson to mark it as read.