Months after releasing GPT-4o, OpenAI has unveiled GPT-4o Mini, its most economical LLM model, at a cost of just 15 cents per million input tokens and 60 cents per million output tokens.
ChatGPT users, including Free, Plus, and Team accounts, will have immediate access to GPT-4o Mini, with Enterprise users gaining access next week.
The model scored 82% on the MMLU benchmark, surpassing its predecessor, GPT-3.5 Turbo. Furthermore, it also outclassed other small models, such as Gemini Flash and Claude Haiku.
“GPT-4o mini is better than other small models at reasoning tasks involving both text and vision, scoring 82.0% on MMLU, a textual intelligence and reasoning benchmark, as compared to 77.9% for Gemini Flash and 73.8% for Claude Haiku,” wrote OpenAI.
GPT-4o Mini’s low cost and high efficiency make it suitable for a wide array of tasks. Its capabilities extend to applications requiring multiple model calls, handling context (like full codebases or conversation histories), and providing real-time text responses. These features are helpful for customer support chatbots and other interactive applications.
Currently, GPT-4o Mini supports text and vision inputs through the API, with plans to include image, video, and audio inputs and outputs in the future. The model boasts a context window of 128K tokens and supports up to 16K output tokens per request, making it highly adaptable for complex and extended interactions.
On the security front, GPT-4o Mini includes the same safety mitigations as GPT-4o, utilising techniques like reinforcement learning with human feedback (RLHF) to enhance accuracy and reliability.
Additionally, the model employs new safety techniques, such as the intrusion hierarchy method, to resist jailbreaks, prompt injections, and system prompt extractions.
“More than 70 external experts in fields like social psychology and misinformation tested GPT-4o to identify potential risks, which we have addressed and plan to share the details of in the forthcoming GPT-4o system card and Preparedness scorecard,” explains OpenAI.
Developers can access GPT-4o Mini through the Assistants API, Chat Completions API, and batch API. It is available at a competitive rate of 15 cents per input token and 60 cents per 1M output tokens, allowing developers to leverage its capabilities at a fraction of the cost of previous models. Fine-tuning for GPT-4o Mini will be rolled out in the coming days.
In May 2024, OpenAI announced ChatGPT Edu for universities at a discounted rate.
In the News: Microsoft rolls out mobile apps for its Designer AI image generator
