• FavTutor
  • Posts
  • New AI model s1 rivals OpenAI’s o1 in under $50 cost

New AI model s1 rivals OpenAI’s o1 in under $50 cost

Including the latest AI news of the week

February 06, 2025

Hello, AI Enthusiasts!

Welcome to FavTutor’s AI Recap! We’ve gathered all the latest and important AI developments for the past 24 hours in one place, just for you.

In Today’s Newsletter: 😀

  • New AI model s1 rivals OpenAI’s o1 in under $50 cost

  • Google Releases New Gemini 2.0 Models

  • TikTok Owner’s New AI makes Lifelike Videos from a Single Photo

OpenAI
😉 New AI model s1 rivals OpenAI’s o1 in under $50 cost

A new AI model called s1, outlined in a research paper, has drawn attention in the AI research community for its cost-effective performance. Stanford and University of Washington researchers demonstrate that powerful test-time scaling for language models can be achieved through remarkably simple methods.

Insights for you:

  • AI researchers were able to train an AI reasoning model with the new model, known as s1. With a single training run lasting 26 minutes, the total computational cost was about US$6.

  • The model performed similarly to cutting-edge reasoning models, such as OpenAI’s o1 and DeepSeek’s R1, on tests measuring math and coding abilities.

  • The s1 model was trained using 16 Nvidia H100 GPUs. They use 1,000 carefully selected training examples while maintaining full transparency and reproducibility.

Seeking impartial news? Meet 1440.

Every day, 3.5 million readers turn to 1440 for their factual news. We sift through 100+ sources to bring you a complete summary of politics, global events, business, and culture, all in a brief 5-minute email. Enjoy an impartial news experience.

Google
⚡️ Google Releases New Gemini 2.0 Models

Google is expanding its AI model family with new Gemini 2.0 variants, each designed for different use cases and offering varying balances of performance and cost. Alongside Gemini 2.0 Flash, Google has launched Gemini 2.0 Pro Experimental for coding performance and complex prompts.

Insights for you:

  • Google is expanding its Gemini AI model family with two new versions of Gemini 2.0: low-cost Flash-Lite is optimized for price/performance, and the experimental Gemini 2.0 Pro targets complex queries and coding.

  • Gemini 2.0 Pro Experimental features a 2 million token context window, the largest in the Gemini lineup.

  • Benchmark data shows that Gemini 2.0 Pro outperforms its predecessors. It scores 91.8% on the MATH benchmark and 44.3% on OpenAI's SimpleQA test.

A
🤳 TikTok Owner’s New AI makes Lifelike Videos from a Single Photo

Researchers at TikTok's parent company ByteDance have presented OmniHuman-1, a new framework for generating videos from image and audio samples. The new system turns still images into videos by adding movement and speech. It is trained on more than 18,700 hours of human video data.

Insights for you:

  • ByteDance has introduced OmniHuman-1, a new framework that can generate high-quality, lifelike animation videos from image and audio examples.

  • OmniHuman processes multiple types of input simultaneously: text, image, audio, and body poses.

  • First, this AI compresses movement data from its various inputs and then refines it by comparing its generated videos to real footage.