🤖 OpenAI just dropped GPT-4o mini

Good morning!

Before we dive into our programming, thank you for filling out the user survey we sent a few days back. We will soon be in touch in regards to the prompt pack.

Additionally, you may notice a few changes in the newsletter. We are undergoing a partnership and re-branding with other publishers in the space. As we go through this transformation, we hope to deliver to you a crisp, informative product that respects your time. The first change is that we will be sending out this email 3x a week to preserve your inbox.

Ok, back to the newsletter.

Today, we’re talking about OpenAI’s GPT-4o mini — a cost-effective and efficient AI model that excels in mathematical reasoning, coding tasks, and multimodal reasoning. It is cheaper and faster than its predecessors and is available for all ChatGPT users and developers via API. 👀

Highlight of the Day

OpenAI has launched GPT-4o mini, their most cost-efficient model to date, aimed at broadening AI accessibility.

Priced at 15 cents per million input tokens and 60 cents per million output tokens, it's significantly cheaper than its predecessors. With an impressive 82% score on the MMLU benchmark, GPT-4o mini excels in text and vision tasks. This model's affordability and versatility are expected to expand AI applications in customer support, data processing, and more. Vineet Khosla, OpenAI’s CTO, states, “We’re committed to driving down costs while enhancing capabilities, making AI more accessible for all.”

GPT-4o mini’s robust performance is evident in various benchmarks. For reasoning tasks, it scored higher than other small models, and it also demonstrated better math and coding proficiency. It outperforms competitors like Gemini Flash and Claude Haiku on academic and multimodal benchmarks, scoring 87.0% in math reasoning (MGSM) and 87.2% in coding performance (HumanEval).

While ChatGPT-4o Mini’s affordability and speed are notable, its success will hinge on maintaining high accuracy and reliability. OpenAI's initiative reflects a broader trend in the AI industry towards making advanced technologies more accessible. Similar efforts by companies like Google and Microsoft have shown that balancing cost, speed, and quality is challenging but achievable.

👉 Read More

Advertise your products to our AI audience

Cool AI Tools

1. Depth
   Analyze user behavior and receive actionable feedback, including UX suggestions and ideas for new features.

   

2. Ragna
   An AI multitool that runs locally on your desktop PC or laptop without an internet connection.

   

3. Kusho
   Generate exhaustive API test suites, then get AI-analyzed results in a single click

   

Best of the Rest

• Proton launches 'privacy-first' AI writing assistant for email that runs on-device

• Google’s Gemini AI will be all over the Paris Olympics broadcast

• Many people think AI is already sentient - and that's a big problem

• Artificial Agency raises a $16M Seed — Behaviour engine for NPCs.

• OpenAI researchers pit two AI models against each other in a game of deception to improve AI explainability

And now your moment of Zen

That’s all for today folks!

• If you’re enjoying the newsletter, share with a friend by sending them this link: 👉 https://www.neonpulse.ai/subscribe

• Looking for past newsletters? You can find them all here.

• Working on a cool A.I. project that you would like us to write about? Reply to this email with details, we’d love to hear from you!