OpenAI has announced GPT-4o mini, a new small model priced to make AI more accessible. The model costs 15 cents per million input tokens and 60 cents per million output tokens, making it 60% cheaper than GPT-3.5 Turbo. This reduction in cost aims to broaden the range of AI applications, from customer support chatbots to processing large volumes of context data.
GPT-4o mini supports both text and vision in its API. Future updates will extend support to image, video, and audio inputs and outputs. The model has a context window of 128K tokens and can handle up to 16K output tokens per request, with knowledge up to October 2023. It also excels in handling non-English text thanks to an improved tokenizer shared with GPT-4o.
The new model performs well across various benchmarks. It scores 82% on the MMLU benchmark for textual intelligence and reasoning, outpacing models like Gemini Flash and Claude Haiku. In math and coding, it achieved 87.0% on MGSM and 87.2% on HumanEval, again surpassing other small models. GPT-4o mini also shows strong performance in multimodal reasoning, scoring 59.4% on the MMMU benchmark.
OpenAI collaborated with companies like Ramp and Superhuman to test GPT-4o mini. These firms found that it outperformed GPT-3.5 Turbo in tasks like extracting structured data from receipts and generating high-quality email responses.
Safety is a priority for GPT-4o mini. The model filters out undesirable content during pre-training and aligns with OpenAI’s policies during post-training using reinforcement learning with human feedback. The model includes built-in safety measures to resist jailbreaks and other prompt attacks, making it reliable and safer for large-scale applications.
The model is available now in the Assistants API, Chat Completions API, and Batch API, with pricing set at 15 cents per million input tokens and 60 cents per million output tokens. GPT-4o mini is accessible to ChatGPT Free, Plus, and Team users starting today, with Enterprise access rolling out next week.
OpenAI aims to reduce AI costs while enhancing capabilities, envisioning a future where AI becomes integral to daily digital experiences. GPT-4o mini sets the stage for more efficient and cost-effective AI development, promising a future where advanced AI is accessible to all.
You can read the original article here: https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/
