Kimi K3.1 Pro lands on Amazon Bedrock
AWS announced on November 6 that Kimi K3.1 Pro — Moonshot AI's open-weight, natively multimodal agentic model with a 1-million-token context window — is now available on Amazon Bedrock, expanding the cloud's open-model portfolio inside its security boundary.
📊 Scorecard: our Bullish call on $AMZN is scored from $259.30, the last close before it ran. Its first result comes after the next close.
The News
AWS announced on November 6 that Kimi K3.1 Pro (Highspeed) is now available on Amazon Bedrock, expanding its open-weight model portfolio. Kimi K3.1 Pro comes from Moonshot AI, a Chinese AI company, and Moonshot describes it as an open-weight, natively multimodal agentic model with a 1-million-token context window — its most capable model in the roughly 3-trillion-parameter class — aimed at agentic coding and document processing inside Bedrock's security boundary. A token is a chunk of text, roughly three-quarters of a word, so a million-token window is about 750,000 words: the model can hold very long documents or codebases in mind at once.
The Bedrock launch is the second Kimi event in two days, and the two should not be confused. On November 4, Moonshot published the full weights of Kimi K3.1 Pro — about 1.56 terabytes across 96 files, with 2.8 trillion total parameters and 104 billion active per token — for anyone to download and run. The weights are the raw material. Bedrock is the finished service: a managed way for businesses to use the model with Amazon's security, privacy and compliance controls, without running it themselves.
Moonshot's own numbers put Kimi K3.1 Pro at 65.7 on Terminal-Bench 4.0, a coding benchmark, against 66.4 for Claude Opus 5.5 and 58.2 for GPT-6 Astra, and at 55 on Artificial Analysis's Intelligence Index at launch (index v4.3.2), three points behind Opus 5.5. Those figures are vendor-reported, not independently verified — Moonshot measured its own model — but the gap to the leading closed models is small, and the model is free to download. That combination is why it is on Bedrock within days of its weights being published.
Kimi K3.1 Pro is now available on Amazon Bedrock — open-weight, natively multimodal, with a 1-million-token context window.
Why It Matters
A model that goes from open weights to a major cloud service in two days is a different kind of event from a model launch. It means the open-model ecosystem now moves at cloud speed: the weights are published, the benchmarks are posted, and within days the model is available inside the security boundary businesses actually use. The 65.7 benchmark score, if it holds up, puts Kimi within a point of Claude Opus 5.5 — and it is free to download.
That is why the verdict is Bullish. For Amazon, every major open model that lands on Bedrock makes the service stickier: businesses that want choice come to the platform with the most models. For the industry, the message is that the closed labs' lead is measured in points, not in years, and the gap is closing in public, on the record, in days. The context window is the other half of the story: one million tokens is about 750,000 words, roughly ten novels, which means the model can work through an entire codebase or document set without forgetting the beginning. That is the capability businesses actually pay for, and it is now available inside Amazon's security boundary for anyone with a Bedrock account.
Analyst's Take
Our call is Bullish. The speed is the story: weights published on November 4, on Amazon's flagship AI service by November 6, with benchmark scores within a point of the best closed model. Open models are no longer a step behind — they are arriving on the biggest cloud platforms in days, not months, and that compresses the advantage the closed labs have lived on.
What to watch: whether businesses actually adopt Kimi on Bedrock, and whether the benchmark scores hold up under independent testing — Moonshot's numbers are vendor-reported. And watch whether the next open release lands on Bedrock even faster. The pattern is now clear: the gap between an open model's release and its arrival on a major cloud is measured in days.
Are you bullish or bearish on this story?
By The Numbers
FAQ
What is Kimi K3.1 Pro?
An AI model from Moonshot AI, a Chinese AI company. It is open-weight — its inner workings are published for anyone to download and run — natively multimodal, meaning it can work with text, images and other inputs together, and built for agentic work: carrying out multi-step tasks, like coding or processing documents, rather than just answering questions.
What does a 1-million-token context window mean?
A token is a chunk of text, roughly three-quarters of a word. The context window is how much text a model can hold in mind at once. One million tokens is about 750,000 words — roughly ten novels — so the model can work through very long documents or codebases without forgetting the beginning.
What is Amazon Bedrock?
Amazon's service for running AI models inside its cloud, with the security, privacy and compliance controls businesses need. Putting Kimi on Bedrock means companies can use the model without running it themselves, and their data stays inside Amazon's security boundary.
How is this different from Moonshot publishing the weights?
Two different events. On November 4, Moonshot published the full weights of Kimi K3.1 Pro — about 1.56 terabytes across 96 files — for anyone to download. On November 6, AWS made the model available on Bedrock, which is the managed, business-ready way to use it. The weights are the raw material; Bedrock is the finished service.
Why does this matter for Amazon?
Because Bedrock's value is the breadth of its model catalogue. Every major open model that lands on Bedrock makes the service more attractive to businesses that want choice — and it puts Amazon in the middle of the open-model ecosystem, even when the model itself comes from a rival lab.
👁
Source: AWS What's New / Moonshot AI (Hugging Face model card)