What I like most about AWS Bedrock is how straightforward it makes building AI applications. I can experiment with different foundation models without having to set up or manage any infrastructure, which saves me a lot of time. I also appreciate how well it integrates with other AWS services, since that makes it easier to put together solutions that are both secure and scalable. Review collected by and hosted on G2.com.

The cost is too expensive, the cold-start latency is an issue, and the model control options feel limited. Review collected by and hosted on G2.com.

What I like most about Bedrock is how it simplifies building AI applications by letting me use different models in one place, without having to deal with complicated setup or ongoing infrastructure management. Review collected by and hosted on G2.com.

One thing I dislike about Bedrock is that some features can feel limited compared with more mature AI platforms, and the pricing can get expensive when you’re using it at a large scale. Review collected by and hosted on G2.com.

What I like most about AWS Bedrock is how it offers easy, unified access to multiple foundation models, which makes it straightforward to build and scale generative AI applications securely. Review collected by and hosted on G2.com.

One minor drawback is that AWS Bedrock’s initial setup and configuration can feel a bit complex at first, although it gets smoother and more manageable as you become familiar with it. Review collected by and hosted on G2.com.

i like about AWS BEDROCK the most is serverless abstraction over multiple, top tier foundation models via single unified API. 1) The "Swap and Play" architecture 2) True enterprise grade privacy by default Review collected by and hosted on G2.com.

The new account throttling nightmare: if you spin brand new AWS Account to

prototype something on bedrock, there are shockingly rare limits like default limits for new accounts are low as 2 -3 requests per minute Review collected by and hosted on G2.com.

What I like best about AWS Bedrock is how easily it lets you access and integrate multiple foundation models through a single managed service. It simplifies building generative AI applications without worrying about infrastructure management. Review collected by and hosted on G2.com.

AWS Bedrock can feel complex for beginners, especially around permissions, networking, and model configurations. Debugging and monitoring model behavior across different providers can also be challenging. Review collected by and hosted on G2.com.

AWS Bedrock makes it easy to work with generative AI by offering access to multiple foundation models without infrastructure overhead. The integration with AWS services is seamless, and scalability is excellent. Review collected by and hosted on G2.com.

Pricing can be tricky, and customization options are still maturing; only limited options can be expected. Since it's only specific to AWS can't be integrated with other cloud providers. Review collected by and hosted on G2.com.

Previous we need to raise a request to enable the bedrock model but now by default all default model are enabled which saves the time and quickly utilise it Review collected by and hosted on G2.com.

Little complexity in use and IAM permission always confuse with permission to bedrock services Review collected by and hosted on G2.com.

We use Claude 3.5 Sonnet model. The biggest advantage of Bedrock is the unified API. Being able to switch between Anthropic Claude 3.5 and Meta Llama 3 just by changing a model ID is a lifesaver for future-proofing. I also love that it’s serverless—I don't have to worry about provisioning GPUs or managing clusters. Plus, since it's inside the AWS ecosystem, the data security and IAM integration make it much easier to get approval from our security team compared to using third-party APIs. Review collected by and hosted on G2.com.

The hidden costs of Knowledge Bases can be annoying; if you aren't careful, the OpenSearch Serverless backend can run up a bill even when you aren't actively using it. Also, model availability is still fragmented across regions—sometimes the newest models are only in US-East-1, which causes latency issues if your main stack is in Europe or Asia Review collected by and hosted on G2.com.

I like that it just works — you pick a model, send a request, and you’re already building something useful. No servers, no headaches, no overthinking. It feels simple, fast, and very “let me get things done. Review collected by and hosted on G2.com.

I dislike that AWS Bedrock can feel a bit heavy. Too many settings, too many steps, and sometimes the pricing feels like a puzzle. It’s powerful, but not always friendly. Review collected by and hosted on G2.com.

We can test different models and switch between them without changing the code or rebuilding our integration. Review collected by and hosted on G2.com.

Using many powerful AI models at the same time on Bedrock can become very costly, especially if you don’t pay attention to how many tokens you use, or if you don’t use caching and avoid unnecessary retries. Review collected by and hosted on G2.com.