"OpenAI competitor" Anthropic is currently in talks for a new round of funding of $3.5 billion, with a valuation reaching $61.5 billion.

According to reports, Anthropic, an artificial intelligence startup supported by Amazon.com, Inc. (AMZN.US), is expected to raise $3.5 billion in its latest round of funding, bringing its total valuation to a staggering $61.5 billion. Sources reveal that investors in the latest round of funding include Lightspeed Venture Partners, General Catalyst, and Bessemer Venture Partners. The startup's recent annual revenue has reached $1.2 billion. Anthropic originally planned to raise $2 billion in this funding round, but the popularity of its chatbot Siasun Robot & Automation Claude evidently helped it secure more funding. Amazon.com, Inc. had previously invested $4 billion in the San Francisco-based Anthropic, including $2.75 billion in March of this year. Other investors include Alphabet Inc. Class C (GOOGL.US) and Salesforce, Inc. (CRM.US). Competitor OpenAI is currently in the midst of a fundraising round, hoping to raise an astonishing $40 billion. If successful, OpenAI's valuation could reach $340 billion. On Monday, Anthropic also announced its new model Claude 3.7 Sonnet. As competition intensifies, Anthropic states that it has developed its "smartest" artificial intelligence model to date. The unique feature of the new model is its so-called hybrid mode, combining reasoning capabilities with the traditional mode of providing real-time answers. Claude 3.7 Sonnet is now available on all CLAUDE plans, Anthropic API, Amazon.com, Inc. Bedrock, and Alphabet Inc. Class C's Vertex AI, with its expanded thinking mode now launched on all platforms except for the free CLAUDE tier. Anthropic co-founder and Chief Scientist Jared Kaplan stated: "This model integrates all functionalities, and we hope to have a coherent artificial intelligence that can help us with everything." Anthropic claims that this is the only "hybrid" model on the market and will be deployed immediately. The new hybrid model provides users with the ability to control how much computing resources are spent on a single question - that is, the time for "reasoning" on that question. Developers can adjust the quantity by sliding a scale, measured by the number of tokens or characters, which the model will process or generate in attempting to predict the answer. By sliding the scale to "0", developers can use the model as a general, non-reasoning artificial intelligence, similar to Microsoft Corporation-supported OpenAI's GPT-4o. OpenAI has a similar feature that allows developers to control the time their reasoning model "thinks". Developers can only set three settings: low, medium, high. However, it is difficult to predict how many tokens the model actually uses at each level, hence predicting the duration of time spent on individual queries.