OpenAI is dedicated to making artificial intelligence widely accessible and releasing GPT-4o mini, this week which will be the most cost-effective small model yet. This new model is set to significantly broaden the scope of AI applications by making advanced intelligence more affordable. GPT-4o mini scores an impressive 82% on MMLU and currently surpasses GPT-4.1 on chat preferences in the LMSYS leaderboard. With pricing at just 15 cents per million input tokens and 60 cents per million output tokens, it is considerably more economical than previous models, being over 60% cheaper than GPT-3.5 Turbo.

Expanding AI Applications with GPT-4o mini

GPT-4o mini’s low cost and latency make it ideal for a variety of tasks, including applications that require multiple model calls (such as chaining or parallelizing APIs), managing large volumes of context (like full code bases or conversation histories), or providing real-time text responses (such as customer support chatbots).

Currently, GPT-4o mini supports text and vision in the API, with future updates to include text, image, video, and audio inputs and outputs. The model boasts a context window of 128K tokens, supports up to 16K output tokens per request, and contains knowledge up to October 2023. The new tokenizer, shared with GPT-4o, enhances cost efficiency in handling non-English text.

Superior Performance in Textual and Multimodal Reasoning

GPT-4o mini outperforms GPT-3.5 Turbo and other small models in academic benchmarks for both textual intelligence and multimodal reasoning. It supports the same range of languages as GPT-4o and excels in function calling, enabling developers to create applications that interact with external systems and handle long-context tasks effectively.

Key Benchmarks Performance

GPT-4o mini has been rigorously evaluated across various benchmarks:

• Reasoning Tasks: Scores 82.0% on MMLU, outperforming Gemini Flash (77.9%) and Claude Haiku (73.8%).

• Math and Coding Proficiency: Excels in mathematical reasoning and coding tasks, scoring 87.0% on MGSM and 87.2% on HumanEval, surpassing Gemini Flash and Claude Haiku.

• Multimodal Reasoning: Scores 59.4% on MMMU, better than Gemini Flash (56.1%) and Claude Haiku (50.2%).

Availability and Pricing

GPT-4o mini is now available as a text and vision model through the Assistants API, Chat Completions API, and Batch API. Pricing is set at 15 cents per million input tokens and 60 cents per million output tokens. Fine-tuning for GPT-4o mini will be available soon.

In ChatGPT, Free, Plus, and Team users can access GPT-4o mini starting today.