Anthropic, the artificial intelligence company behind Claude AI, has unveiled its latest family of large language models called Claude 3.
Anthropic says the new Claude 3 family of AI models performs as well as or better than leading models from Google and OpenAI. Unlike earlier versions, Claude 3 is also multimodal, able to understand text and photo inputs.
The Claude 3 Lineup
The Claude 3 lineup consists of three powerful AI models – Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus, with Opus being their flagship “most intelligent model.”
Also Read: OpenAI reveals Sora, its text to video AI model
According to Anthropic, Claude 3 will answer more questions, understand longer instructions, and be more accurate. Claude 3 can understand more context, meaning it can process more information.
Enhanced Contextual Understanding
A key advantage of Claude 3 is its enhanced ability to handle nuanced prompts that may have been inappropriately rejected by older models due to over-restrictive filters, allowing for more natural interactions.
Previous versions of Claude refused to answer some prompts that were harmless, which the company writes “suggests a lack of contextual understanding.” The new models are less likely to refuse to answer prompts that toe the line of their safety guardrails.
In addition to traditional language tasks, the multimodal nature of Claude 3 enables it to process a variety of data inputs, such as images, charts, and research papers, with blazing speed.
Anthropic claims Claude 3 models can give near-instant results even while parsing dense material like a research paper.
A blog post says Haiku, the smallest version of Claude 3, is “the fastest and most cost-effective model on the market,” able to read a dense research paper complete with charts and graphs “in less than three seconds.””
Benchmarking Superiority
The performance of Claude 3, especially the top-tier Opus model, has been nothing short of impressive in rigorous benchmarking tests conducted by Anthropic.
Not only did Opus outperform cutting-edge models like OpenAI’s GPT-4 on graduate-level reasoning tasks, but it also demonstrated superior capabilities in areas like math problem-solving, coding, and logical reasoning.
Also Read: Gemini vs. ChatGPT: How does Google’s latest AI offering compare?
Even compared to Anthropic’s own previous model, Claude 2.1, the new Claude 3 versions exhibit substantial performance gains, with the mid-tier Sonnet model being twice as fast, making it ideal for time-sensitive applications.
Powerful Training Process
To develop these cutting-edge AI capabilities, Anthropic trained the Claude 3 models on a massive and diverse dataset combining proprietary data sources with publicly available information as of August 2023.
This intensive training process leveraged the powerful cloud computing infrastructure of Amazon Web Services (AWS) and Google Cloud – two tech giants that have substantially invested in Anthropic, with Amazon alone providing $4 billion in funding.
Availability and Platforms
The Claude 3 lineup will be accessible through multiple channels. Anthropic’s own API will offer all three models – Haiku, Sonnet, and Opus. Additionally, the models will be available on AWS’s Bedrock model library and Google’s Vertex AI platform, thanks to the investments from these cloud giants.
For users of the free Claude AI service, the current version runs on the mid-tier Claude 3 Sonnet model as of this writing. However, to experience the full capabilities of Anthropic’s flagship Claude 3 Opus model, users will need to subscribe to the company’s Pro tier plan at $20 per month. This paid offering unlocks access to Opus, touted as Anthropic’s “most intelligent” large language model to date, with top performance across various benchmarks.
The Future of Multimodal AI
With its multimodal inputs, state-of-the-art performance surpassing industry leaders, and enhanced safety capabilities, Anthropic’s Claude 3 family of AI models is poised to push the boundaries of what’s possible in natural language AI.
Developers and businesses can soon harness these powerful models for applications like chatbots, autocomplete, data extraction, and more.
