The Switzerland eroticfloodgates have opened for building AI reasoning models on the cheap.
Researchers at Stanford and the University of Washington have developed a model that performs comparably to OpenAI o1 and DeepSeek R1 models in math and coding — for less than $50 of cloud compute credits.
What's more, the model was trained on only 1,000 questions, and took just 26 minutes and 16 Nvidia H100 GPUs. Stanford researcher Niklas Muennighoff said in a email to Mashable that the cost is an estimate based on the GPU runtime and number of H100 GPUs used.
The AI industry of late is all about how new approaches to the pre and post training process can massively save computing costs, as evidenced by DeepSeek's disruptive impact. On top of that, developers are now able to build on top of existing AI models at little or no cost, through APIs, open-source access, and even closed-source models by distilling their data, bringing the costs down even more.
According to the team's research paper which was published last Friday, s1 was trained on a dataset consisting of "1,000 carefully curated questions paired with reasoning traces and answers distilled from Gemini Thinking Experimental." Google's Gemini Thinking Experimental model is accessible with daily limits through AI Studio. While it's a closed-source model, that clearly hasn't stopped researchers from making use of its responses.
SEE ALSO: OpenAI launches 'deep research' AI agent for ChatGPTNext, the researchers used an "off the shelf" pretrained model from Alibaba-owned lab, Qwen, and performed supervised fine-tuning of its curated dataset. Then, the team created a token budget to control the amount of compute time for testing the model. If s1 went over budget on thinking tokens, it was cut off and forced to generate whatever answer it came up with. If the researchers wanted the model to spend more "test-time compute" on a problem, they would simply tell the model to "wait," which extended its thinking time and led to more accurate results.
By controlling the amount of time and compute spent on a problem, the researchers were able to show how increased thinking team leads to improved performance.
S1 is one example of open-source reasoning models that have been developed for a fraction of the cost of flagship models from Google and OpenAI. In January, UC Berkeley researchers released an open-source reasoning model called Sky-T1 that cost $450, "demonstrating that it is possible to replicate high-level reasoning capabilities affordably and efficiently," per its blog post. There's also the open-source rStar-Math reasoning model from Microsoft Asia researchers, Tulu 3 from non profit research institute Ai2, and HuggingFace has its own initiative to replicate DeepSeek's R1.
As high-quality models become more accessible and cheaper, we're starting to see a power shift from the few AI heavy hitters, to the many.
Topics Artificial Intelligence OpenAI
Previous：My Enemies Defeated Me / For Nothing
Next：Mine over Matter
Watch NASA bring Artemis home with a spectacular splashdown and recovery
Recapping Dante: Canto 9, or Dial V for Virgil by Alexander Aciman
Nail Art by Sadie Stein
Into the Woods by Sadie Stein
Death in the Mango Orchard
Amateur Night by Adam Wilson
Unconscious by Sadie Stein
C. S. Lewis Reviews The Hobbit
After Eden
Smut by Sadie Stein
The Singing Left
C. S. Lewis Reviews The Hobbit
Modern Austen, and Other News by Sadie Stein
Actual Size by Sadie Stein
Fun, Games, and Extractivism
The Literary World by Sadie Stein
Happy Belated by Sadie Stein
What We Talk About When We Talk About Ill
Hand of War
How the Magic Happens, and Other News by Sadie Stein
Best Garmin deal: Save $50 on Garmin Lily 2iPhone 16e vs iPhone 16: What are the differences?Logitech G305 LIGHTSPEED gaming mouse deal: 40% off at AmazonBest Eufy robot vacuum deal: Get the Omni C20 for $250 offChase Bank may block Zelle payments to social media contacts as scams surgeBest Roomba deal: Roomba Combo j5 hits new recordNYT mini crossword today: Answers for February 19, 2025Best earbuds deal: Save $50 on Beats Studio BudsBest Garmin deal: Save $50 on Garmin Lily 2Best speaker deal: Take 30% off the Ultimate Ears Wonderboom 4Dortmund vs. Sporting 2025 livestream: Watch Champions League for freeHow to speed up an Instagram ReelWordle today: The answer and hints for February 18, 2025NYT Connections hints and answers for February 19: Tips to solve 'Connections' #619.Bayern Munich vs. Celtic 2025 livestream: Watch Champions League for freeBest Garmin deal: Save $50 on Garmin Lily 2Best earbuds deal: Save $50 on Beats Studio BudsBest laptop deal: Save 48% on the Samsung Galaxy Book4 Pro at AmazonGet the Samsung Galaxy Tab S9 FE for 34% off at AmazonNYT Connections hints and answers for February 18: Tips to solve 'Connections' #618. NYT Connections Sports Edition hints and answers for February 1: Tips to solve Connections #131 Scotland vs. Italy 2025 livestream: Watch Six Nations for free Best Valentine's Day deal: Amazon Fresh has BOGO Valentine's Day candy Elon Musk's DOGE takeover is reportedly being spearheaded by young college grads Los Angeles Lakers vs. New York Knicks 2025 livestream: Watch NBA online First porn app 'approved' for the iPhone in Europe. Apple isn't happy. NYT Strands hints, answers for February 4 USAID website offline as Musk carries out federal bloodletting Jesse Eisenberg used ChatGPT to understand his anxiety over ordering a bagel NYT Connections hints and answers for February 1: Tips to solve 'Connections' #601. Wordle today: The answer and hints for February 2, 2025 Best Beats Studio Pro deal: $179.99 at Amazon and Best Buy Mark Zuckerberg removed tampons from men's restrooms. Meta employees put them back. Dallas Mavericks vs. Philadelphia 76ers 2025 livestream: Watch NBA online Best mesh WiFi deal: Save $340 on eero Max 7 mesh WiFi system Wordle today: The answer and hints for February 4, 2025 Best Amazon Fire TV Stick Lite deal: $17.99 at Amazon Best headphones deal: Save $100 on the Beats Solo 4 Best Meta Quest 3S deal: Save $50.99 at Amazon Super Bowl LIX livestreams: How to watch Chiefs vs Eagles for free without cable
2.036s , 8215.484375 kb
Copyright © 2025 Powered by 【Switzerland erotic】,Pursuit Information Network