Amazon Bedrock adds an India inference option for Kimi K3

An AWS machine-learning blog post dated 3 November 2026 says Amazon Bedrock now supports in-country inference in India for Kimi K3, in addition to global and US-based inference. It says the India geographic profile, with model id in.moonshotai.kimi-k3, is for when inference processing must stay within the Mumbai and Hyderabad Regions. It says the global profile id is global.moonshotai.kimi-k3. It says data is encrypted in transit between Regions and stays on the AWS network.

The post quotes Aditya Patawari, VP Engineering at Ultrahuman, on an assistant called Jade. The quotation says the assistant answers questions using a person's Ring and glucose data, across more than 14,000 conversations a day, and that with prompt caching the team has cut LLM costs by 80 percent and made multi-turn chats 85 percent faster. The quotation says Mumbai is a fallback Region and that health data stays inside AWS.

The post says logs and metrics are stored in the source Region where the request was submitted, regardless of which destination Region processed it. It says model invocation logging is opt-in and can contain prompts and outputs. It does not state a new token price for the India profile.

What it means for business: a workload that must keep Kimi K3 inference inside India can use the India profile the post names. The global profile is a different option. The 80 percent and 85 percent figures are in a customer quotation in the AWS post, not an AWS price cut. This describes the post. It is not advice on a residency rule.