OpenAI’s GPT-4o Mini: Check Features, Capabilities and Pricing

On July 18, 2024, OpenAI announced the release of its latest artificial intelligence (AI) model, GPT-4o Mini. The company says that it is their most cost-efficient small model yet. This new model, said to be faster than any other AI model of OpenAI, is currently being released for developers. ChatGPT’s web and mobile application users will be able to access it as well. 

GPT-4o Mini promises to reform the range of applications built with AI by significantly reducing costs while maintaining high-performance standards. It scores 82% on MMLU and outperforms GPT-41 on chat preferences in the LMSYS leaderboard. The Mini model outperforms other industry-leading small AI models as well. 

Let’s take a look at its features. 

GPT-4o Mini Features

Here are some of the key features of GPT-4o Mini:

1. High Performance and Accuracy

According to OpenAI’s official blog post, GPT-4o Mini scored 82% on the Massive Multitask Language Understanding (MMLU) benchmark. It outperformed its predecessor, GPT-3.5 Turbo, and other small models such as Gemini Flash and Claude Haiku. Also, it excels in different domains, like reasoning, mathematical proficiency, and coding.

2. Multimodal Support:

Currently supporting text and vision inputs, GPT-4o Mini will eventually allow image, video, and audio inputs. Developers will be able to design a wide variety of apps, from sophisticated data analysis tools to chatbots for customer service.

3. Large Context Window:

The mini model has a context window of 128,000 tokens and can support up to 16,000 output tokens per request. It can handle extensive input data, which is ideal for applications requiring large volumes of context, such as processing full code bases or maintaining conversation history.

4. Enhanced Multimodal Reasoning:

GPT-4o Mini performs well in multimodal reasoning problems, scoring  59.4% on the MMMU benchmark. This capacity makes it ideal for applications that require the integration of text and visual data.

5. Math and coding proficiency

The model excels at mathematical reasoning and coding skills, beating prior small models on the market. On the MGSM, which measures math reasoning, GPT-4o mini scored 87.0%, compared to 75.5% for Gemini Flash and 71.7% for Claude Haiku. GPT-4o tiny earned 87.2% on HumanEval, a coding performance test, compared to 71.5% for Gemini Flash and 75.9% for Claude Haiku.

Perplexity AI is releasing its Pro Search function for intricate queries on Android Devices

Capabilities

The GPT-4o Mini has been carefully assessed against several major benchmarks. Here are its performance benchmarks: 

• Reasoning Tasks: 82.0% on MMLU, outperformed Gemini Flash (77.9%) and Claude Haiku (73.8%).
• Math Proficiency: 87.0% on MGSM, compared to 75.5% for Gemini Flash and 71.7% for Claude Haiku.
• Coding Performance: 87.2% on HumanEval, surpassed Gemini Flash (71.5%) and Claude Haiku (75.9%).
• Multimodal Reasoning: 59.4% on MMMU, scored higher than Gemini Flash (56.1%) and Claude Haiku (50.2%).

Open AI Search Engine: How this AI-Powered Search Product is Different from Google? Check Here

Pricing and Accessibility

GPT-4o mini is available as a text and vision model via the Assistants API, Chat Completions API, and Batch API. It is priced at 15 cents per million input tokens and 60 cents per million output tokens, (approximately equal to 2500 pages in a standard book). It is more affordable than previous models and over 60% cheaper than GPT-3.5 Turbo. 

ChatGPT Free, Plus, and Team members can gain access to GPT-4o mini instead of GPT-3.5. However, enterprise users will gain access in the coming week. 

What is OpenAI’s Strawberry? A secret project for AI Model Deep Research and Reasoning

The Bottom Line

Open AI says, “We envision a future where models become seamlessly integrated in every app and on every website. GPT-4o mini is paving the way for developers to build and scale powerful AI applications more efficiently and affordably.”

The release of GPT-4o Mini marks a big step forward in AI development, combining high performance with cost. With its improved performance, a wide variety of functions, and enhanced safety features, GPT-4o Mini is poised to lead the next generation of AI innovation, making advanced intelligence more accessible, dependable, and integrated into our daily digital lives. 

Microsoft AI Chief Mustafa Suleyman Praises OpenAI’s Sam Altman, Discusses Future of AI and Regulation