OpenAI GPT-OSS 120B (Open-Weight)

OpenAI offers several model families: the cutting-edge GPT-4o series (including the flagship GPT-4o and the more powerful GPT-4o Plus, plus the cost-effective GPT-4o Mini) for advanced multimodal tasks, the specialized o1 series (o1-Preview, o1-Mini) for state-of-the-art reasoning, and older models like GPT-3.5 Turbo for efficient text generation. They also provide models for image generation (DALL-E 3) and other specific tasks.

OpenAI continues to charge per 1,000 tokens, with distinct rates for input (prompt) and output (completion) tokens. More advanced models like GPT-4o Plus and the o1 series generally have higher per-token costs, reflecting their enhanced capabilities. Cached input pricing is often available at a discount (typically 50%) for repeated prompts.

The GPT-4o series (GPT-4o, GPT-4o Plus, GPT-4o Mini) are OpenAI's primary multimodal models, excelling at natively handling text, audio, images, and even video inputs/outputs with impressive speed and efficiency. The o1 series (o1-Preview, o1-Mini) is more specialized, focusing on achieving new frontiers in complex reasoning, mathematical abilities, and logical deduction, often at a higher price point for those specific capabilities.

For the latest generation models like GPT-4o Plus, a 256K token context window is becoming standard. Other models like GPT-4o and the o1 series typically offer 128K token context windows. This allows for processing very large amounts of information in a single interaction.