OpenAI is making AI more accessible with GPT-4o mini, a new cost-effective small language model that offers superior performance in text and code compared to previous options, all at a reduced price. The high cost of AI models has previously been a barrier for many developers, but GPT-4o mini breaks this barrier with a price tag of just 15 cents per million input tokens and 60 cents per million output tokens.
Benchmarks reveal that GPT-4o mini outperforms other small models in crucial areas like textual reasoning, mathematical problem-solving, and coding proficiency. This versatility extends to multimodal tasks, where the model can handle scenarios involving both text and images.
The language model is trained with meticulous filtering to remove harmful content, and further fine-tuned using reinforcement learning to ensure accurate and reliable responses. Additionally, a novel instruction hierarchy method safeguards against misuse, making GPT-4o mini a trustworthy tool for developers.
Currently, GPT-4o mini is available through OpenAI’s APIs for text and vision tasks. To further empower developers, OpenAI plans to introduce fine-tuning capabilities in the near future, allowing for even greater control over the model’s outputs.
 |
| GPT-4o mini |