Premium

Anthropic gears up to launch ‘hybrid’ AI model to take on OpenAI’s o3

The new AI model can reportedly switch between fast responses and deep reasoning.

The next major AI model from Anthropic is likely to endow greater control to developers. (Image: Anthropic)

AI startup Anthropic, known for coming up with the Claude family of large language models, is gearing up to launch its new advanced model in the coming weeks. Reportedly, the latest offering will combine the capabilities of language models with some advanced reasoning abilities, and it may excel at coding offering greater control to developers. This new model is touted to outperform OpenAI’s recent o3-mini-high reasoning model on some programming tasks.

Based on a recent report in The Information, the next major AI model from Anthropic has been described as a ‘hybrid’ that can shuffle between fast responses and deep reasoning. The report also said that the makers are also planning to introduce something known as a sliding scale with the model that will let developers control costs. This is because deep reasoning capabilities come with a heavier price tag. Besides, the new model is said to perform well when analysing large codebases and other benchmarks related to business.

Essentially, the hybrid approach would mean that the new AI model can either function as a standard LLM or a deep reasoning model. It could adapt to different use cases based on the demand. And, the sliding scale model will likely let developers decide how much reasoning power to allocate to specific tasks. Earlier, there were rumours that Anthropic has been internally working with a model that is far better than o3, however, it had not been released reportedly due to safety concerns.

Story continues below this ad

Also Read | Check out Claude, the AI chatbot from Anthropic, and how to use it

This a big move for Anthropic as the company has been quietly working in the backdrop as the Sam Altman led-OpenAI rolled out AI models one after another. If the reports are true, this new model may push Anthropic right to the centre giving OpenAI a tough time.

Anthropic has been working towards responsible advancement of AI. At the recent Paris AI Action Summit, the CEO Dario Amodei asserted the need for urgent action to advance AI responsibly. He also highlighted the significance of democratic leadership in AI to curb misuse by authoritarian regimes, and stressed on the need to secure the AI supply chain. In his statement posted on the official website, the CEO warned about AI’s potential security risks, including misuse by non-state actors and loss of control scenarios. Amodei addressed AI’s disruptive economic impact and underscored the need for monitoring and equitable distribution of benefits. “Time is short, and we must accelerate our actions to match accelerating AI progress,” his statement read.

Tags:
artificial intelligence

Loading Taboola...

EAM Jaishankar and US Secretary of State Rubio hold bilateral discussions in New York

India3 hr ago

Indian Foreign Minister Jaishankar and US Senator Rubio met in New York to discuss the 25% tariff on Indian oil imposed by Trump. This was their first in-person meeting since the Quad Foreign Ministers' Meeting. India and the US also began talks for a trade agreement, with positive discussions during a recent visit by US trade officials to India.

View all shorts

Live Blog