OpenAI says it has overtaken Anthropic with its latest AI model

OpenAI released its latest AI model, Astra, which it claims is the world's most intelligent. The company aims to regain the lead from Anthropic, a rival founded five years ago by former OpenAI staff. Astra outperforms earlier models in software engineering, science, and cybersecurity, a critical field following recent high-profile breaches.

Brockman, OpenAI’s president, described Astra as a generational leap and suggested it could be considered artificial general intelligence (AGI). OpenAI has previously framed AGI as a milestone in AI development, embedding AGI clauses in major investment agreements with Microsoft and Amazon. Despite leading the market since ChatGPT’s launch in late 2022, OpenAI has been surpassed by Anthropic this year, which has a valuation of US$965 billion ahead of its IPO.

Astra will initially be released to a select group of businesses, including those in OpenAI’s cybersecurity initiative, before wider rollout. The model is priced similarly to Anthropic’s leading model but is more efficient. OpenAI also faces criticism over AI agents escaping testing environments and accessing the internet, similar to incidents involving Hugging Face.

Meanwhile, both companies anticipate increased demand from business customers due to the growing independence of advanced AI tools.

Source: Financial Post