OpenAI launches GPT-4o mini, a smaller and cheaper AI model that promises enhanced performance. The company announced on Thursday that this new model is cheaper and faster than its current top-tier AI models. It is available for developers and through the ChatGPT web and mobile app starting today. Enterprise users will gain access next week.
According to OpenAI, GPT-4o mini excels in reasoning tasks involving both text and vision. Small AI models like GPT-4o mini are gaining popularity among developers due to their speed and cost advantages over larger models, such as GPT-4 Omni or Claude 3.5 Sonnet. These models are ideal for high-volume, simple tasks that require repeated AI model usage.
GPT-4o mini is set to replace GPT-3.5 Turbo as OpenAI’s smallest model. The company reports that GPT-4o mini scores 82% on the MMLU benchmark for reasoning, compared to 79% for Gemini 1.5 Flash and 75% for Claude 3 Haiku, based on data from Artificial Analysis. On the MGSM benchmark for math reasoning, GPT-4o mini scored 87%, outperforming Flash at 78% and Haiku at 72%.
Affordability and Future Capabilities
OpenAI claims the GPT-4o mini is more affordable to run than its previous models, being over 60% cheaper than the GPT-3.5 Turbo. Currently, GPT-4o mini supports text and vision in the API, with plans to add video and audio capabilities in the future.
OpenAI’s head of Product API, Olivier Godement, emphasized the importance of affordability in making AI accessible worldwide. He described GPT-4o mini as a significant step towards this goal. For developers using OpenAI’s API, GPT-4o mini is priced at 15 cents per million input tokens and 60 cents per million output tokens. The model features a context window of 128,000 tokens, approximately the length of a book, and a knowledge cutoff of October 2023.
Speed and Performance
Developers are excited as OpenAI launches GPT-4o mini, a smaller and cheaper AI model suitable for high-volume tasks. George Cameron, Co-Founder of Artificial Analysis, highlighted the speed of GPT-4o mini. He noted that it has a median output speed of 202 tokens per second, which is more than twice as fast as GPT-4o and GPT-3.5 Turbo. This makes the GPT-4o mini a compelling choice for speed-dependent applications, including many consumer applications and agentic approaches to using large language models (LLMs).
On the same day, OpenAI also announced new tools for enterprise customers. These tools include the Enterprise Compliance API, designed to help businesses in regulated industries like finance, healthcare, legal services, and government comply with logging and audit requirements.
Additionally, OpenAI is offering more granular control for workspace GPTs. Previously, admins could only fully allow or block GPT actions in their workspace. Now, they can create an approved list of domains that GPTs can interact with.
Analysis of OpenAI’s GPT-4o Mini
OpenAI’s GPT-4o mini is a notable development in the AI industry. It promises high performance and lower costs, making advanced AI more accessible. The model scores 82% on the MMLU benchmark and 87% on MGSM, suggesting it can handle complex text and math reasoning tasks well.
As OpenAI launches GPT-4o mini, a smaller and cheaper AI model, businesses anticipate significant cost savings. Affordability is a key advantage of GPT-4o mini. Being over 60% cheaper than GPT-3.5 Turbo, it provides a cost-effective solution for businesses, especially smaller companies and startups. This could democratize AI, allowing more organizations to benefit from advanced AI capabilities.
OpenAI’s plans to add video and audio processing to GPT-4o mini are exciting. These features could expand the model’s applications, making it useful in customer service. However, OpenAI has not disclosed the exact size of the GPT-4o mini, which limits transparency. Users might find that the model’s efficiency varies depending on the specific tasks and data it encounters. Real-world performance can differ from benchmark results, so practical applications should be monitored closely.
Also Read: OpenAI’s Aggressive Court Tactics Test First Amendment Limits, Sparking Legal Debate.