The latest update to Google DeepMind’s top Gemini AI model introduces a dial that allows developers to control the level of reasoning the system applies to responses. This feature aims to save money for developers by addressing the issue of overthinking in reasoning models, which can increase costs and energy consumption. Reasoning models, a focus of AI development since late last year, can cost up to $200 to complete a single task. Outputs from these models are about six times more expensive to generate when reasoning is enabled. The shift towards reasoning models reflects a change in the AI industry, moving away from scaling laws and towards longer thinking times for better responses. AI companies now spend more on inferencing than training, a trend expected to grow with the rise of reasoning models. This increased inferencing also contributes to a larger environmental footprint. Despite the dominance of reasoning models, alternatives like the open-weight DeepSeek model, which caused a nearly $1 trillion stock market dip upon its announcement, offer cheaper options. However, for tasks requiring high accuracy and precision, such as coding, math, and finance, proprietary models may still be preferred.
Haider.
@slow_developer
·
Apr 11 “reasoning models are dirt cheap compared to humans”OpenAI Research Scientist, Noam Brown:the reasoning models may seem costly compared to GPT-4o…but they are more cost-effective than human experts, especially as they start to surpass human performance in certain areas.
Brian Chau (SF April 15th-20th)
@psychosort
·
Apr 16 They way they measure compute cost is even more insane – counting cumulative compute cost of a *company* instead of the compute cost of a model. This means a company that trained twenty 5.1 million dollar models would be liable under the law.
Colin Fraser
@colin_fraser
·
Mar 6 Like the $200/month OpenAI Deep Research agent can’t figure out which NBA players are on which teams with better than about 15% accuracy but in 2 years it’s supposed to be a billion times smarter than Albert Einstein? Seems hard to believe.
Jason Whaling
@Jason_Whaling
·
Apr 11 AI was built on a lie.If OpenAI had announced:“We’re scraping your work, training an AI on it, and charging $200/month for access…”People would’ve lost their minds. But that’s not what happened.
Anil Kumar Tulsiram
@Anil_Tulsiram
·
Apr 17 Many individuals continue to rely on the free version and then criticize AI for providing inaccurate or suboptimal responses. As full-time investors, it is essential that we recognize the value of investing in our own operations. If we are unwilling to make such investments
To provide the best experiences, we use technologies like cookies to store and/or access device information. Consenting to these technologies will allow us to process data such as browsing behavior or unique IDs on this site. Not consenting or withdrawing consent, may adversely affect certain features and functions.
Functional
Always active
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
Preferences
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
Statistics
The technical storage or access that is used exclusively for statistical purposes.The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Marketing
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
