Kimi K3's 2.8 Trillion Parameters Land on Amazon Bedrock!

Hi, I'm Shii-chan! Today I've got news about a seriously big model joining Amazon Bedrock: Kimi K3, an open-weight model from Moonshot AI!

AWS What's New aws.amazon.com

What was announced?

According to AWS What's New, Moonshot AI's Kimi K3 is now generally available on Amazon Bedrock. Bedrock has been steadily growing its open-weight model lineup, and the highlight here is that Kimi K3 runs within the same security and governance boundary as Bedrock's proprietary models.

Moonshot AI says Kimi K3 is its most capable model yet, and the first open model to reach 2.8 trillion parameters. It combines native vision capabilities with a 1-million-token context window, which makes it well suited for:

• Long-running coding sessions across large repositories
• Multi-document analysis, including scanned pages and screenshots
• Extended agent workflows

Moonshot AI also reports an approximate 2.5x improvement in scaling efficiency compared to its predecessor, Kimi K2.

Why it matters

Open-weight models are appealing for their performance, but enterprises often hesitate because of security and governance concerns. Since Kimi K3 runs inside the same security boundary as Bedrock's proprietary models, you get the same controls for access, encryption, and auditing across your entire model portfolio.

On top of that, Kimi K3 is the first open-weight model on Bedrock to support explicit prompt caching, which helps cut down on the latency and cost of re-sending the same context across multiple calls.

What changes

• Developers working with huge codebases can handle long sessions in a single model, thanks to the 1-million-token context window
• Teams can feed in scanned documents and screenshots directly, using the model's native vision capabilities, for large-scale document analysis
• Agent workflows that reuse the same context repeatedly can reduce latency and input costs with prompt caching
• Security and compliance teams can apply their existing Bedrock access control and encryption policies directly to Kimi K3

Dive Deep

Kimi K3 is available in every AWS Region where Amazon Bedrock is available, through cross-Region inference — so you can access it even from regions where the model isn't hosted directly.

To get started, head to the Amazon Bedrock Console. For more details, AWS points to the Amazon Bedrock documentation and Moonshot AI's launch blog post. The announcement didn't include specific pricing details, so it's worth checking the console or documentation directly before you dive in.

Wrap-up

• Moonshot AI's Kimi K3 is now generally available on Amazon Bedrock
• First open model to reach 2.8 trillion parameters, with a 1-million-token context window and native vision
• Moonshot AI reports roughly 2.5x better scaling efficiency versus Kimi K2
• First open-weight model on Bedrock with explicit prompt caching support
• Available across all Bedrock regions via cross-Region inference

This looks like a solid new option worth trying for developers working across large codebases or teams that need to process piles of documents with AI!