March 4, 2013 - Business Insider reported this evening BST thatAmazonAn artificial intelligence model with advanced "reasoning" capabilities is being developed.
According to a person with direct knowledge of the project, the new product is scheduled to be released as early as June as "Nova" brand launch, Nova is a series of generative AI models that Amazon launched late last year.
The person said that Amazon wants the new model to use an"Mixed reasoning" approachBothGiving quick answersand can be performed in the same systemMore complex thinking.
inference modelIt has become the next breakthrough in artificial intelligence in recent years. These types of models are slower to compute, but can solve more complex problems through multiple attempts and chained thinking back. Companies like Google, OpenAI, and Anthropic have recently introduced their own inference models, and DeepSeek has attracted a lot of attention for its more efficient approach.
One of Amazon's primary goals, people familiar with the matter said, is to get its Nova inference model on the priceOutperforms other programs including OpenAI's o1, Anthropic's Claude 3.7 Sonnet, and Google's Gemini 2.0 Flash Thinking. and other competitors.
Previously, Amazon has stated that its existing Nova model is more reliable than third-party models available through the Bedrock platformCheap at least 75%.
Additionally, Amazon plans to make its upcoming inference modelingTop 5 in external benchmarkingThese benchmarks assess software development and math skills and cover tests such as SWE, Berkeley Function Call Ranking, and AIME.
It also means Amazon is in more direct competition with AI startup Anthropic. The latter just recently released its latest inference model, Claude 3.7 Sonnet, which uses a similar hybrid inference approach, combiningQuick answers and longer chains of thought output.
It's worth noting that Amazon has invested $8 billion in Anthropic (note: currently around Rs. 58,369 crore) and the two companies have been working very closely together in areas such as AI chips and cloud computing.
