OpenAI launches GPT-4o mini, a smaller and more economical AI model, designed to increase accessibility and affordability of artificial intelligence. This model is priced significantly lower than its predecessors, such as GPT-3.5 Turbo, costing a mere 15 cents per million input tokens and 60 cents per million output tokens.
GPT-4o Mini: Performance and Capabilities
According to OpenAI’s official blog, GPT-4o mini exhibits impressive performance, achieving 82% on the MMLU benchmark and outperforming GPT-41 on the LMSYS leaderboard for chat preferences. The model’s affordability and quick response times make it well-suited for tasks requiring multiple model calls, extensive context handling, or real-time text interactions, such as customer support chatbots.
Currently, GPT-4o mini supports text and vision inputs, with plans to expand to image, video, and audio inputs and outputs in the future. It boasts a 128K token context window, supports up to 16K output tokens per request, and possesses knowledge up to October 2023. Additionally, its enhanced tokenizer improves the cost-effectiveness of handling non-English text.
GPT-4o Mini Excels in Diverse Applications
The model demonstrates exceptional performance in both academic and practical domains, surpassing other small models in reasoning, math, and coding tasks. It achieved 87% in mathematical reasoning and 87.2% in coding performance on benchmarks like MGSM and HumanEval, respectively.
Safety Measures Implemented in GPT-4o Mini
OpenAI has incorporated robust safety measures into GPT-4o mini. These measures include filtering out harmful content during pre-training and utilizing reinforcement learning with human feedback (RLHF) to align the coding performance with safety policies. Moreover, over 70 external experts have evaluated the model to identify and mitigate potential risks.
Availability and Future Enhancements
GPT-4o mini is now accessible through the Assistants API, Chat Completions API, and Batch API. It will be available to Free, Plus, and Team users on ChatGPT starting today, with enterprise users gaining access next week. OpenAI also intends to introduce fine-tuning capabilities for GPT-4o mini soon.
OpenAI’s Vision for Accessible AI
OpenAI’s mission is to continually reduce costs while enhancing AI capabilities. The launch of GPT-4o mini represents a significant step towards integrating powerful AI into everyday applications, making advanced intelligence more accessible and affordable for developers and users alike.
Add Comment