OpenAI releases free AI model 'GPT-4o mini', supports multi-modal image and voice input
On July 18, 2024, OpenAI announced the cost-effective, compact model 'GPT-4o mini'. Despite its low cost, the GPT-4o mini has outperformed GPT-4 in some benchmarks and is now available to free users and ChatGPT Plus users as the successor to GPT-3.5.
GPT-4o mini: advancing cost-efficient intelligence | OpenAI
OpenAI Releases GPT-4o Mini, a Cheaper Version of Flagship AI Model - Bloomberg
https://www.bloomberg.com/news/articles/2024-07-18/openai-releases-gpt-4o-mini-a-cheaper-version-of-flagship-ai-model
OpenAI launches GPT-4o mini, which will replace GPT-3.5 in ChatGPT | Ars Technica
https://arstechnica.com/information-technology/2024/07/openai-launches-gpt-4o-mini-which-will-replace-gpt-3-5-in-chatgpt/
The GPT-4o mini announced by OpenAI achieved a score of 82% in the MMLU benchmark, which measures the multitasking language understanding of language models, and was rated higher than GPT-4 on the LMSYS leaderboard.
The results of each benchmark are as follows. Although the GPT-4o mini did not reach the GPT-4o, it outperformed GPT-3.5 Turbo, Google's lightweight model 'Gemini Flash', and Anthropic's small model 'Claude Haiku' in most benchmarks.
Despite its excellent performance, it is low-cost, costing just 15 cents per million input tokens and 60 cents per million output tokens, more than 60% cheaper than GPT-3.5 Turbo.
Introducing GPT-4o mini! It's our most intelligent and affordable small model, available today in the API. GPT-4o mini is significantly smarter and cheaper than GPT-3.5 Turbo.
https://t.co/sqJsFEYHWq pic.twitter.com/g6jMttp1mF — OpenAI Developers (@OpenAIDevs) July 18, 2024
GPT-4o mini has an input context window size of 128,000 tokens, supports up to 16,000 output tokens per request, and has knowledge up to October 2023.
Aside from the cost, the new AI model is the first from OpenAI to incorporate a security technique called 'instruction hierarchy,' which prioritizes instructions, making the model less susceptible to prompt injection attacks and jailbreaks.
The GPT-4o mini already supports image input via an API, and OpenAI plans to support text, image, video, and audio input and output in the future.
ChatGPT users on the Free, Plus, and Team plans can use GPT-4o mini instead of GPT-3.5, and Enterprise users will soon have access to GPT-4o mini.
Olivier Gordman, head of API products at OpenAI, said, 'Our mission is to enable state-of-the-art technology and build the most powerful and useful applications, so we will continue to develop cutting-edge models and push the boundaries, but we also want to provide the best small models.
Related Posts:
in Software, Posted by log1l_ks