OpenAI announced the new AI model GPT-4o mini, a smaller version of the GPT-4

OpenAI has announced a new compact AI model, the GPT-4o mini, which makes GPT-4 technologies more accessible.

Here's What We Know

The model costs 15 cents per 1 million input tokens and 60 cents per 1 million output tokens to use, which is significantly cheaper than GPT-3.5 Turbo. The GPT-4o mini outperforms other models such as Gemini Flash, Claude Haiku, and GPT-3.5 Turbo in benchmarks and almost reaches the level of GPT-4o.

The model focuses on textual intelligence and multimodal reasoning, including mathematical reasoning and coding tasks. It has a context window of 128 thousand tokens and knowledge until October 2023. Support for images, video and audio is planned for the future.

The rollout of GPT-4o mini begins today, 19 July and it replaces GPT-3.5 in the Free, Plus and Team subscriptions for end users. Enterprise customers will have access to the model starting next week.

Source: OpenAI