Anthropic CEO Dario Amodei has confirmed that the company still plans to release Claude 3.5 Opus, their most powerful large language model (LLM). This news, revealed on the Lex Fridman Podcast, comes after Anthropic's July release of Claude 3.5 Sonnet and Claude 3.5 Haiku, leaving the AI community wondering about the future of Opus.
Amodei explained that Anthropic's strategy has been to focus on improving the capabilities of their smaller models first. He highlighted the significant improvements in Sonnet 3.5, which now performs comparably to Opus 3 at the same cost as the previous Sonnet 3. Even Haiku 3.5, the smallest model, has reached the performance level of Opus 3.
“Not giving you an exact date, but as far as we know, the plan is still to have a Claude 3.5 Opus,” said Amodei.
Reason for the Delay
Amodei attributed the delay in Opus 3.5 to the intensive computational demands of pre-training, extensive safety testing, and the company's ambition to create a model that addresses new challenges and provides exceptional value.
"That (pre-training) uses, these days, tens of thousands, sometimes many tens of thousands of GPUs or TPUs...often training for months," Amodei explained.
Recent Feature Releases
Despite the delay in Opus 3.5, Anthropic has been actively releasing new features. The "Computer Use" feature has generated considerable excitement, potentially revolutionizing human-computer interaction. Another recent addition is the "Visual PDF" tool, enabling Claude to analyze PDFs and extract insights from images, charts, and graphs.
The Upcoming LLM Battle
With OpenAI's impending o1 release and Google's Gemini 2.0 on the horizon, the competition in the LLM space is heating up. The arrival of Claude 3.5 Opus promises to be a significant contender in this evolving landscape.
