AWS adds Kimi K3.1 Pro Highspeed model to Amazon Bedrock

Amazon Web Services published a blog post on November 10, 2026 introducing Kimi K3.1 Pro Highspeed, an open model from Moonshot AI, on Amazon Bedrock. According to AWS's Kimi K3.1 Pro Highspeed announcement, the new version generates output 2 to 3 times faster than the prior Kimi K3.1 Pro while keeping a 1 million token context window and processing data within the AWS boundary.

What is changing

• A 1 million token context window.
• Output generation 2 to 3 times faster than the previous Kimi K3.1 Pro.
• Three pricing tiers: Priority, Standard, and Flex, with Flex priced at half the Standard rate.
• Three cross-Region inference profiles: global, US-only, and India-only.
• Support for Chat Completions, Responses, Converse, and Invoke APIs, plus tool calling, structured output, image input, and streaming.

What it means for builders

For teams building on Bedrock, Highspeed offers a faster and cheaper open-weight alternative to US frontier-lab models for coding and agentic workloads. AWS positions the model for long-horizon agentic tasks, financial modeling, office documents, and coding tasks including game engines and CAD. Lower latency combined with tiered pricing, especially the discounted Flex tier, gives builders more room to run longer agent sessions without raising costs.

Data handling

AWS states that customer data processed through Kimi K3.1 Pro Highspeed stays within the AWS data boundary, is not shared with Moonshot AI, and is not used to train the model, with zero data retention. That guarantee gives builders a data residency commitment alongside the open-weight model itself, rather than relying only on the provider's own policies.

Sources

• Introducing Kimi K3.1 Pro on Amazon Bedrock (Amazon Web Services, November 10, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.