GPT-4o is the latest version of OpenAI's GPT-4 large language model, released in May 2024. It introduces several key improvements over previous versions:
- 2x faster at generating tokens than GPT-4 Turbo[1]
- 50% cheaper pricing at $5 per million input tokens and $15 per million output tokens[1]
- 5x higher rate limits up to 10 million tokens per minute[1]
- Enhanced vision capabilities across most tasks[1]
- Improved non-English language capabilities with a new tokenizer for more efficient non-English text processing[1]
- 128K context window and a knowledge cut-off date of October 2023[1][3]
- Video understanding support by converting videos to frames (2-4 frames per second) for input[1]
- Improved translation abilities for smoother cross-language communication[1]
GPT-4o maintains GPT-4's strong performance on text, reasoning, and coding tasks, while adding significant multimodal capabilities to process image and video inputs[2][3]. However, like previous GPT models, it can still produce incorrect information, biased outputs, and unsafe responses in some cases[2][3].
GPT-4o is available through the GPT-4 API for developers, with pricing of $5 per million input tokens and $15 per million output tokens[1][4]. It has a context window of 128,000 tokens and a throughput of 90.4 tokens per second, making it a high-performance model[4].
Comments
Post a Comment