Moonshot AI's Kimi K3.1 Pro lands on Amazon Bedrock with OpenAI-compatible APIs

AWS added Moonshot AI's Kimi K3.1 Pro to Amazon Bedrock. The model is reachable through three interfaces: OpenAI-compatible Chat Completions, the Responses API, and Bedrock's own Converse API. That lets teams already built on OpenAI SDKs switch with minimal code changes. AWS's launch post walks through explicit prompt caching for cheaper repeated context and shows the model working as a coding assistant inside OpenCode.

A companion Kimi K3.1 Pro Highspeed variant brings a 1M-token context window, according to Unite.ai coverage. It joins a Bedrock lineup that gained Claude Fable 5.2 and OpenAI's Daybreak models in the same week. AWS's pitch is clear: Bedrock aims to be the neutral marketplace for frontier models, whether from US labs or Chinese open-weight challengers.

This matters for cost-sensitive agent work. Chinese open-weight models such as Kimi, DeepSeek and Qwen have competed on price, and developers are actively debating self-hosting Kimi K3.1 Pro against paying for API access. Hosting it on Bedrock gives enterprises a managed option with AWS governance, without running GPUs themselves.

Watch whether enterprise procurement teams accept a Chinese-developed model running inside AWS. Also watch how its Bedrock pricing compares with Gemini 4 Carbon, which is setting a low bar for frontier-grade agentic work this week.