Highlights:

  • The most recent release of Claude 3.5 Sonnet achieved a score of 49% on SWE-bench Verified, a benchmark assessing neural networks’ programming abilities.
  • Anthropic develops the Claude series of large language models, releasing the advanced Claude 3.5 Sonnet in June, followed by an upgraded version in August that can perform tasks in applications for users.

Multiple recent reports indicate that a competitor to OpenAI, Anthropic PBC, aims to raise up to USD two billion in new funding.

Sources cited by CNBC and the Wall Street Journal suggest that the investment could value Anthropic at USD 60 billion—over three times its valuation from last year’s funding round led by Menlo Ventures. Lightspeed Venture Partners is reportedly spearheading the company’s latest fundraising efforts.

Anthropic creates a series of large language models known as Claude. In June, it launched the most advanced model in the series, Claude 3.5 Sonnet. Two months later, the company introduced an enhanced version of the LLM capable of performing actions within applications on behalf of users.

The newest release of Claude 3.5 Sonnet achieved a 49% score on SWE-bench Verified, a benchmark evaluating neural networks’ programming abilities—marking a nearly 16% improvement over the initial version launched in June. Alongside it, Anthropic offers Claude 3.5 Haiku, a less advanced but more cost-effective model, which scored 40.6% on the same benchmark.

Anthropic provides its LLMs via an application programming interface and offers a ChatGPT-like chatbot called Claude Enterprise. This service can generate code, create sales forecasts, and handle other business tasks using datasets supplied by users.

According to a source cited by CNBC, Anthropic currently generates USD 875 million in annualized recurring revenue, primarily from enterprise deals. In contrast, sources reported in September that OpenAI is on pace to close 2024 with USD 3.7 billion in revenue and is projected to reach USD 11.6 billion this year. CNBC’s source did not provide details on Anthropic’s growth rate.

Upon releasing the latest version of Claude 3.5 Sonnet, Anthropic announced plans to introduce a more advanced model, Claude 3.5 Opus, in the future. The new funding it is seeking could potentially aid in advancing the development efforts.

OpenAI’s latest reasoning models leverage a machine learning technique called test-time compute, which enhances an LLM’s response quality by utilizing more hardware to process prompts. However, this approach also increases inference costs. If Anthropic opts to adopt test-time compute in its future LLMs, the USD two billion it seeks to raise could help offset the associated expenses.

This funding round follows three prior multibillion-dollar investments.

In 2023, Anthropic received an investment of up to USD 2 billion from Google LLC. Later that year, Amazon.com Inc. secured a USD four billion stake in the LLM developer. In November, AWS doubled its investment through an agreement that also established it as Anthropic’s primary partner for AI training infrastructure.

Last year, OpenAI completed a USD 6.6 billion funding round at a valuation of USD 157 million. It also secured a USD four billion line of credit from a consortium of banks. Cohere Inc., another key player in the general-purpose LLM space, previously raised USD 500 million from Advanced Micro Devices Inc. and other investors.