The eroticized childhood traumafloodgates 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
Melania Trump bought her own dress for Republican National ConventionViola Davis regrets starring in 'The Help' for deeply personal reasonsHoly hell, the 512GB iPhone XS Max costs $1,449Here's all the milestones Tim Cook boasted about at the 2018 Apple EventA Harry Potter star is joining 'Dancing With the Stars'Mark Zuckerberg subtly made a case for not breaking up FacebookIt's Britain's hottest day of the year and people simply can't copeJournalists report being forced from convention floor after 'Never Trump' protestsHere are the leaked colors and storage sizes of new iPhones XS and XRWatch this musician expertly twerk and play the flute at the same timeSelena Gomez criticized after weighing in on the Taylor Swift/Kanye West dramaSan Francisco rolls out safety campaign to ensure riders get in the right Uber or LyftLittle baseball fan experiences a whirlwind of emotions during 18How to watch the RNC if you're a cord cutterThis probably fake app gets other people to pick up your dog's poopMark Zuckerberg explains how Facebook is fighting election meddling10 video games we can't wait to escape with this fallMelania Trump bought her own dress for Republican National ConventionNintendo is now worth more than Sony thanks to 'Pokémon Go'Someone approved 'white elevators' signs at Republican convention Best luggage deal: Get up to 60% off on spring break luggage at Amazon All the best signs from Climate Marches around the world How to have a successful double or group date Nature to greet People's Climate March with record heat TikTok and Instagram diet tips to avoid Lyft expands Women+ Connect safety feature nationwide Tesla Model Y refresh might not be happening this year Nothing to launch its next smartphone on March 5 Slack is about to TL;DR your lengthy work threads Tinder, Bumble, Hinge: It's easy to catfish strangers on dating apps 'Abbott Elementary's talking heads have meaning all of their own 'Boycott Tesla' ads aired during Super Bowl 2024 spotlight self Wordle today: The answer and hints for February 14 OpenAI terminates accounts of confirmed state India's rural solar revolution hasn't delivered on its promise Donald Trump's very own staff member handed him fake news. And Trump believed it. Elon Musk and U.S. tech giants tell Trump not to ditch the Paris Climate Agreement Here's how I feel about all this Stephen Hawking 'news' going around Two adorable, newly discovered Yoda How to watch 'The Color Purple' — release date, Max subscription deals
2.9103s , 10192.4765625 kb
Copyright © 2025 Powered by 【eroticized childhood trauma】,Unobstructed Information Network