Kimi K3.1 Pro Highspeed on Amazon Bedrock
Moonshot's Kimi K3.1 Pro Highspeed, a post-trained variant that generates output 2-3x faster with improved reasoning, is now available on Amazon Bedrock with data residency options including India-only.
Why it matters
Kimi's 2.8T open-weight flagship reaches AWS enterprise buyers with zero data retention and India-only inference.
What you could build with it
An Indian startup could run Kimi K3.1 Pro Highspeed on Bedrock's India-only inference profile to power a long-document agent for contracts and filings, keeping data in-country while using the 1M-token context to read whole files in one pass.
Does it hold up?
Usable today on Bedrock for AWS teams; the model itself is proven on the hosted Kimi API, though self-hosting the 1.56TB weights remains a multi-GPU job.
Built with Kimi K3.1 Pro Highspeed on Amazon Bedrock
- Moonshot AI releases Kimi K3.1 Pro weights: 2.8T parameter MoE, 1M token context, and Gated DSA sparse attentionmlllm · Technical rundown of the weights release: 2.8T total, 104B active, 1M context, Gated DSA in 24 of 93 layers, vendor benchmarks.
- Kimi K3.1 Prodsebastien · Independent overview covering the two flavors, pricing, open weights on Hugging Face since November 4, and the HN community reaction.
- Kimi K3.1 Pro an Open-Weight 2.8-Trillion-Parameter Modelmedium · Deep dive noting the strong agentic subtests against a weaker overall intelligence index, and the missing training-compute disclosure.
- vllm-project/vllmgithub · Kimi K3.1 Pro ships with published vLLM deployment recipes for multi-GPU serving.
- sgl-project/sglanggithub · SGLang deployment recipes for Kimi K3.1 Pro were published alongside the weights.
Learn more
First spotted on article: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.