Unlocking Innovation: AWS and Anthropic Collaborate to Expand the Possibilities of Generative AI

Amazon Bedrock is the top destination for constructing and expanding creative AI applications with extensive language models (LLM) and other fundamental models (FMs). It allows users to utilize a range of high-performing FMs, like the Claude family of models by Anthropic, to create personalized creative AI applications. Reflecting on 2021, when Anthropic initially began developing on AWS, it was unimaginable how revolutionary the Claude family of models would become. Through Amazon Bedrock, we have been democratizing cutting-edge creative AI models for businesses of all magnitudes. In just a few months since Amazon Bedrock was made widely available on September 28, 2023, over 10K customers have been utilizing it for deployment, with many leveraging Claude. Companies such as ADP, Broadridge, Cloudera, Dana-Farber Cancer Institute, Genesys, Genomics England, GoDaddy, Intuit, M1 Finance, Perplexity AI, Proto Hologram, Rocket Companies, and others are utilizing Anthropic’s Claude models on Amazon Bedrock to foster innovation in creative AI and construct transformative customer interactions. Today, we are thrilled to announce a significant milestone with the introduction of the next generation of Claude coming to Amazon Bedrock: Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku.

Unveiling Anthropic’s Claude 3 models

Anthropic is introducing its advanced generation of Claude with three optimized models for distinct applications. Haiku is the quickest and most cost-efficient model available, ideal for near-instant responses. Sonnet, for the majority of workloads, is twice as fast as Claude 2 and Claude 2.1, with heightened intelligence levels. It excels in cognitive tasks that demand prompt responses, such as information retrieval or sales automation. It strikes a perfect balance between intelligence and speed, crucial for enterprise applications. Opus is the most advanced, state-of-the-art FM with deep analytical, mathematical, and coding capabilities, offering exceptional performance on highly intricate tasks. It can handle open-ended prompts and unexplored scenarios with remarkable ease, including task automation, hypothesis formulation, and analysis of visual data. Moreover, Sonnet is now accessible on Amazon Bedrock. Initial evaluations by Anthropic suggest that the Claude 3 model family surpasses comparable models in mathematical problem-solving and multilingual mathematical benchmark assessments, essential benchmarks for modern LLMs.

  1. Vision capabilities – Claude 3 models have been trained to interpret structured and unstructured data across various formats, including not only text but also visuals, charts, diagrams, and more. This enables businesses to create creative AI applications integrating diverse multimedia sources and addressing cross-domain challenges. For instance, pharmaceutical companies can analyze drug research reports alongside protein structure illustrations to expedite discoveries. Media entities can automatically produce image descriptions or video scripts.
  2. Leading benchmarks – Claude 3 outperforms existing models in standardized assessments like math problems, coding exercises, and scientific reasoning. Customers can refine domain-specific experimental processes in manufacturing or validate financial reports based on contextual data efficiently and accurately using AI-enabled solutions.

    Source: https://www.anthropic.com/news/claude-3-family

    Specifically, Opus excels in most common evaluation benchmarks for AI systems, including expertise at the undergraduate level (MMLU), expert reasoning at the graduate level (GPQA), basic mathematics (GSM8K), and more. It demonstrates high comprehension and fluency in complex tasks, spearheading general intelligence advancements.

  3. Reduced error rates – Businesses demand accurate, predictable outputs from AI systems guiding automated processes or customer engagements. Claude 3 models mitigate errors through constitutional AI methods that offer insight into the model’s decision-making process and enhance precision. Claude 3 Opus showcases an estimated 2x improvement in accuracy compared to Claude 2.1 on challenging open-ended queries, diminishing the chance of inaccurate responses. As corporate clients rely on Claude in sectors like healthcare, finance, and legal research, minimizing errors is crucial for safety and effectiveness. The Claude 3 family establishes a new standard for dependable creative AI output.

Advantages of Anthropic Claude 3 FMs on Amazon Bedrock

Through Amazon Bedrock, users will have convenient access to utilize Anthropic’s latest models, encompassing not only natural language models but also a broader spectrum of multimodal AI models capable of advanced reasoning incorporating text, visuals, charts, and more. Our partnership has already accelerated the adoption of creative AI for users and delivered tangible value to them. Here are a few examples of how customers are leveraging Anthropic’s Claude models on Amazon Bedrock:

“We are developing an AI solution on AWS to assist customers in planning remarkable trips and crafting life-altering experiences with tailored travel schedules. By utilizing Claude on Amazon Bedrock, we slashed itinerary generation expenses by nearly 80% when we swiftly established a scalable, secure AI platform capable of organizing our content in minutes to provide cohesive, extremely accurate travel tips. Now, we can modify and personalize our content in various manners on our digital platforms, catering to customer preferences, all the while highlighting trusted local insights– in a similar vein to Lonely Planet’s legacy of 50 years.”

— Chris Whyde, Senior VP of Engineering and Data Science, Lonely Planet

“We are collaborating with AWS and Anthropic to host our customized, finely-tuned Anthropic Claude model on Amazon Bedrock to support our strategy of swiftly deploying creative AI solutions at scale, with state-of-the-art encryption, data protection, and secure AI technology embedded throughout our operations. Our novel Lexis+ AI platform incorporates conversational search, insightful summarization, and intelligent legal drafting features, empowering legal professionals to enhance their efficiency, effectiveness, and productivity.”

— Jeff Reihl, Executive VP and CTO, LexisNexis Legal & Professional

“At Broadridge, we are streamlining the understanding of regulatory reporting mandates to enhance transparency and efficiency for our clients in local and global financial markets. By leveraging Claude on Amazon Bedrock, we are achieving superior accuracy in our experiments with processing and summarization capabilities. Amazon Bedrock offers us a selection of LLMs, and we appreciate its performance and integration capabilities.”

— Saumin Patel, VP Engineering generative AI, Broadridge

The Claude 3 model family caters to diverse requirements, allowing clients to select the most suitable model for their specific needs, crucial for developing successful prototypes and operational systems delivering genuine impact—whether for a novel product, feature, or process that enhances the bottom line. Keeping customer needs at the forefront, Anthropic and AWS are delivering precisely where it counts for businesses of all scales:

  1. Enhanced efficiency – Claude 3 models substantially accelerate real-time engagements due to optimizations across hardware and software.
  2. Enhanced precision and dependability – Leveraging extensive scaling and novel self-supervision techniques, expected improvements of 2x in accuracy for complex questions over extended contexts translate to AI that is even more supportive, secure, and reliable.
  3. Simplified and secured customization – Customization features like retrieval-augmented generation (RAG) simplify model training with proprietary data and constructing applications supported by diverse data sources, ensuring AI tailored to individual requirements. Furthermore, proprietary data is never exposed to public networks, stays within the AWS environment, securely transfers through VPC, and remains encrypted during transit and storage.

AWS and Anthropic are continuously upholding our commitment to advancing creative AI responsibly. By continuously enhancing model capabilities and adhering to frameworks like Constitutional AI or the White House voluntary AI commitments, we accelerate the secure, ethical development, and deployment of this groundbreaking technology.

The future of creative AI

In the future, clients will construct entirely new categories of creative AI-driven applications and experiences utilizing the latest model generation. We have barely scratched the surface of creative AI’s potential to automate intricate processes, enhance human expertise, and reshape digital interactions. Expect unparalleled levels of innovation as customers adopt Anthropic’s models enhanced with multimodal capabilities, utilizing all the necessary tools to construct and expand creative AI applications on Amazon Bedrock. Envisage sophisticated conversational aides delivering rapid and contextually rich responses, visualize personalized recommendation engines seamlessly incorporating relevant visuals, schematics, and associated knowledge to intuitively guide decisions. Envision scientific exploration accelerated by creative AI that can comprehend experiments, formulate hypotheses, and even suggest uncharted research areas. The potentialities that can be realized by fully leveraging creative AI through Amazon Bedrock are endless. Our partnership ensures that businesses and pioneers globally will have the resources to reach the next pinnacle of creative AI-powered innovation responsibly and for the common good.


We are still in the early stages of creative AI, but robust collaboration and a focus on innovation are ushering in a new era of creative AI on AWS. We are excited to see the groundbreaking advancements that customers will develop next.


Explore the following resources to delve deeper into this announcement:

About the author

Swami Sivasubramanian is Vice President of Data and Machine Learning at AWS. In this capacity, Swami supervises all AWS Database, Analytics, and AI & Machine Learning services. His team’s objective is to aid organizations in leveraging their data comprehensively with an end-to-end data solution encompassing storage, access, analysis, visualization, and prediction.

Leave a Reply

Your email address will not be published. Required fields are marked *