SyncAI.news, a Varaisys broadcasting
ChatGPT 6.2 Astra: How OpenAI's new AI model could help legal professionals; internet seeks 'EU certification'
MA

Mint: AI

· 1 min read

IndiaMint: AI

ChatGPT 6.2 Astra: How OpenAI's new AI model could help legal professionals; internet seeks 'EU certification'

OpenAI launched GPT-6.2 Astra for demanding professional tasks and GPT-6.2 Sol for everyday workflows. The company says the models can handle more complex tasks with fewer output tokens, lowering costs. Here's how it will help lawyers.

OpenAI has introduced two new AI models, GPT-6.2 Astra and GPT-6.2 Sol, featuring enhanced performance for workplace and everyday applications. The latest releases are also designed to lower the expenses associated with deploying advanced AI systems, particularly for businesses and developers.

Netizens have sought the “EU certification” for this model.

According to the company, the models can execute more sophisticated tasks while generating fewer output tokens, helping users get more done without exhausting their usage allowances. The move is intended to make powerful AI capabilities more accessible and economical for organisations integrating the technology into their workflows.

GPT-6.2 Astra targets demanding professional requirements, such as creating documents, presentations and spreadsheets, developing software and conducting scientific research. GPT-6.2 Sol, on the other hand, focuses on regular user needs and everyday workflows that require completing several steps.

OpenAI said Astra delivered a stronger performance than its predecessor in the Agents' Last Exam benchmark, an evaluation designed to measure an AI model's ability to complete complex professional assignments involving multiple stages.

The new model achieved a score 5.2 points above GPT-6 Astra while bringing its operating cost down to less than a third of its predecessor's. OpenAI also claimed that Astra outperformed competitor Claude Fable 5.2 by 5.5 points, despite costing just 5% as much to operate.

Internet reacts

Reactions to the announcement were mixed, with some users poking fun at the model's presentation and appearance, while others expressed cautious optimism.

Garvit Bhirani

Original source

This story was published by Mint: AI. SyncAI.news shows a preview; the complete article is on the publisher's site.

Read the full story on livemint.com

Similar News