OpenAI
The company behind GPT-4, ChatGPT, and DALL-E
OpenAI builds the most widely used large language models in the world. If your product has AI features, there's a decent chance it's powered by OpenAI under the hood. Their API is basically the default starting point for any team building AI into their product.
openai.comUse Cases
Text generation
Writing assistance, content creation, chatbots, summarization — anything that involves generating or manipulating text.
Code generation
Powering coding assistants like GitHub Copilot and Cursor. Turns out, LLMs are really good at code.
Image generation
DALL-E for creating images from text descriptions. GPT-4V for understanding images.
Embeddings
Turning text into vectors for search and similarity matching. The backbone of most RAG systems.
Key Features
- →GPT-4o — their most capable model for most tasks
- →GPT-4o mini — cheaper, faster, good enough for many use cases
- →Assistants API — build AI agents with tools and file access
- →Fine-tuning — customize models on your data
- →Function calling — let the model trigger actions in your app
Why This Matters for PMs
As a PM, you need to understand OpenAI's model lineup because your engineering team will ask you to make tradeoffs: GPT-4o is better but costs more and is slower. GPT-4o mini is cheaper and faster but less capable. The right choice depends on your use case, your users' expectations, and your budget. You should also understand rate limits, token pricing, and the fact that OpenAI can change their models at any time.