Making AI Headlines · Page 3
OpenAI's Mid-Range Model Makes the Flagship Explain Its Expense Account
· New York time
OpenAI has released GPT-6.1 Terra, a mid-range model priced at half its predecessor's API cost. Analysts say it offers a strong amount of capability for the money, particularly for routine computing tasks and AI-driven computer control. It is not OpenAI's largest model, but it is gaining ground on bigger offerings. For developers, lower usage prices can change the assessment of which tasks are economical to automate. A feature used occasionally has a different cost profile from one invoked on every customer request. A less expensive model that performs well on common work may therefore attract applications that do not require the company's most powerful system. The release also complicates the industry's favorite ranking contest. The biggest model may still win many demanding comparisons, but a buyer choosing a model for routine work has another question: how much capability is needed for this particular job? That is an awkward conversation for the flagship, which arrived expecting to be celebrated as the smartest person in the room. Terra is happy to sit at the less glamorous desk and handle the ordinary requests at a lower price. In the office economy, the prestige model may discover that writing an agenda is not the best argument for its expense account.
Read this story in the full newspaper →Satire. Not factual reporting. Read the original reporting from the linked publishers.