Advanced AI models generate up to 50 times more CO₂ emissions than more common LLMs when answering the same questions
When you buy through links on our articles, Future and its syndication partners may earn a commission.
The more accurate we try to make AI models, the bigger their carbon footprint — with some prompts producing up to 50 times more carbon dioxide emissions than others, a new study has revealed.
Reasoning models, such as Anthropic's Claude, OpenAI's o3 and DeepSeek's R1, are specialized large language models (LLMs) that dedicate more time and computing power to produce more accurate responses than their predecessors.
Yet, aside from some impressive results, these models have been shown to face severe limitations in their ability to crack complex problems. Now, a team of researchers has highlighted another constraint on the models' performance — their exorbitant carbon footprint. They published their findings June 19 in the journal Frontiers in Communication.
"The environmental impact of questioning trained LLMs is strongly determined by their reasoning approach, with explicit reasoning processes significantly driving up energy consumption and carbon emissions," study first author Maximilian Dauner, a researcher at Hochschule München University of Applied Sciences in Germany, said in a statement. "We found that reasoning-enabled models produced up to 50 times more CO₂ emissions than concise response models."
To answer the prompts given to them, LLMs break up language into tokens — word chunks that are converted into a string of numbers before being fed into neural networks. These neural networks are tuned using training data that calculates the probabilities of certain patterns appearing. They then use these probabilities to generate responses.
Reasoning models further attempt to boost accuracy using a process known as "chain-of-thought." This is a technique that works by breaking down one complex problem into smaller, more digestible intermediary steps that follow a logical flow, mimicking how humans might arrive at the conclusion to the same problem.
Related: AI 'hallucinates' constantly, but there's a solution
However, these models have significantly higher energy demands than conventional LLMs, posing a potential economic bottleneck for companies and users wishing to deploy them. Yet, despite some research into the environmental impacts of growing AI adoption more generally, comparisons between the carbon footprints of different models remain relatively rare.
To examine the CO₂ emissions produced by different models, the scientists behind the new study asked 14 LLMs 1,000 questions across different topics. The different models had between 7 and 72 billion parameters.
The computations were performed using a Perun framework (which analyzes LLM performance and the energy it requires) on an NVIDIA A100 GPU. The team then converted energy usage into CO₂ by assuming each kilowatt-hour of energy produces 480 grams of CO₂.
Their results show that, on average, reasoning models generated 543.5 tokens per question compared to just 37.7 tokens for more concise models. These extra tokens — amounting to more computations — meant that the more accurate reasoning models produced more CO₂.
The most accurate model was the 72 billion parameter Cogito model, which answered 84.9% of the benchmark questions correctly. Cogito released three times the CO₂ emissions of similarly sized models made to generate answers more concisely.
"Currently, we see a clear accuracy-sustainability trade-off inherent in LLM technologies," said Dauner. "None of the models that kept emissions below 500 grams of CO₂ equivalent [total greenhouse gases released] achieved higher than 80% accuracy on answering the 1,000 questions correctly."
RELATED STORIES
—Replika AI chatbot is sexually harassing users, including minors, new study claims
—OpenAI's 'smartest' AI model was explicitly told to shut down — and it refused
—AI benchmarking platform is helping top companies rig their model performances, study claims
But the issues go beyond accuracy. Questions that needed longer reasoning times, like in algebra or philosophy, caused emissions to spike six times higher than straightforward look-up queries.
The researchers' calculations also show that the emissions depended on the models that were chosen. To answer 60,000 questions, DeepSeek's 70 billion parameter R1 model would produce the CO₂ emitted by a round-trip flight between New York and London. Alibaba Cloud's 72 billion parameter Qwen 2.5 model, however, would be able to answer these with similar accuracy rates for a third of the emissions.
The study's findings aren't definitive; emissions may vary depending on the hardware used and the energy grids used to supply their power, the researchers emphasized. But they should prompt AI users to think before they deploy the technology, the researchers noted.
"If users know the exact CO₂ cost of their AI-generated outputs, such as casually turning themselves into an action figure, they might be more selective and thoughtful about when and how they use these technologies," Dauner said.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
29 minutes ago
- Yahoo
EIB to allot 70 billion euros for tech sector in 2025-2027-officials
BRUSSELS (Reuters) -The European Investment Bank is likely to announce on Friday plans to pump 70 billion euros into the development of European technology firms over the next three years, EU officials said. The programme, called Tech EU, is meant to help Europe compete with China and the United States in the race for innovative clean and digital technologies. The EIB, the biggest multilateral lender in the world with a balance sheet total of 556 billion euros, expects its own 70 bln euros to mobilise a further 250 billion euros of private cash as investors crowd into projects supported by the EIB, EU officials said. The 70 billion is to be split into 20 billion euros for equity and quasi-equity, 40 billion euros for loans and 10 billion for guarantees in 2025-2027, the officials said. The plan is to complement European Commission efforts to support higher risk ventures and innovative companies throughout their investment journey, from proof of concept to an initial public offering. The EIB wants to focus on supercomputing, artificial intelligence, digital infrastructure, critical raw materials, green industries such as offshore wind, health, security and defence technologies, robotics and advanced materials, the officials said. Sign in to access your portfolio
Yahoo
32 minutes ago
- Yahoo
B. Riley Lowers Price Target on FuelCell Energy, Inc. (FCEL)
FuelCell Energy, Inc. (NASDAQ:FCEL) is among the 13 Best Hydrogen and Fuel Cell Stocks to Buy According to Analysts. Riley has maintained its Neutral rating on FuelCell Energy, Inc. (NASDAQ:FCEL) and reduced its price target from $9 to $8 in response to the company's fiscal Q2 results. An industrial setting, with a fuel cell power plant against a backdrop of smoke stacks. The company mentioned that its recent cost-cutting initiatives resulted in a slight decrease in forecasts. FuelCell has reduced its headcount by 22% and plans to reduce annualized cost by 30% in comparison to fiscal 2024. The company's move toward stricter cost controls in the face of persistent operational difficulties is reflected in these initiatives. B. Riley's updated forecast reflects cautious investor sentiment as FuelCell Energy, Inc. (NASDAQ:FCEL) works through its reorganization. FuelCell Energy, Inc. (NASDAQ:FCEL) is a fuel cell power firm and one of the best hydrogen stocks. The company creates, manufactures, sells, installs, operates, and services fuel cell products and electrolysis platforms that reduce carbon emissions and generate hydrogen. FuelCell Energy, Inc. (NASDAQ:FCEL) provides services to many industries, including commercial and hospitality, wastewater treatment, education and healthcare, data centers, and industrial. Geographically, the business is active in Europe, Canada, South Korea, and the United States. The USA and South Korea account for the majority of revenue. While we acknowledge the potential of FCEL as an investment, we believe certain AI stocks offer greater upside potential and carry less downside risk. If you're looking for an extremely undervalued AI stock that also stands to benefit significantly from Trump-era tariffs and the onshoring trend, see our free report on the best short-term AI stock. READ NEXT: 10 High-Growth EV Stocks to Invest In and 13 Best Car Stocks to Buy in 2025. Disclosure. None. Sign in to access your portfolio
Yahoo
32 minutes ago
- Yahoo
Telegram boss says he has fathered more than 100 children
The multi-billionaire founder of instant messaging app Telegram, Pavel Durov, says he has fathered more than 100 children. "The clinic, where I started donating sperm 15 years ago to help a friend, told me that more than 100 babies had been conceived this way in 12 countries," Mr Durov told French political magazine Le Point. Mr Durov, who says he is the "official father" of six other children with three different partners, added that all of his offspring will share his estimated $13.9bn (£10.3bn) fortune. He also reiterated that he denies any wrongdoing in connection with serious criminal charges he faces in France. "They are all my children and will all have the same rights! I don't want them to tear each other apart after my death," Mr Durov said. But the self-exiled Russian technology tycoon told the magazine that none of his children would have access to their inheritance for 30 years. "I want them to live like normal people, to build themselves up alone, to learn to trust themselves, to be able to create, not to be dependent on a bank account," he said. The 40-year-old said he had written a will now because his job "involves risks – defending freedoms earns you many enemies, including within powerful states". His app, Telegram, known for its focus on privacy and encrypted messaging, has more than a billion monthly active users. Mr Durov also addressed criminal charges he faces in France, where he was arrested last year after being accused of failing to properly moderate the app to reduce criminality. He has denied failing to cooperate with law enforcement over drug trafficking, child sexual abuse content and fraud. Telegram has previously denied having insufficient moderation. In the Le Point interview he described the charges as "totally absurd". "Just because criminals use our messaging service among many others doesn't make those who run it criminals," he added.
