Originally published on ConvertElementor: https://convertelementor.com/blog/moonshot-just-put-a-2-8-trillion-parameter-model-on-hugging-face-kimi-k3-1-pro-s-open-weights-are-real-and-the-fine-print-matters-november-2026-20261105
Moonshot AI published the full weights of Kimi K3.1 Pro on Hugging Face on November 4, 2026 — 2.8 trillion parameters, a 1-million-token context window, and a license that is genuinely open for most of us but not for everyone.
What actually landed
Moonshot posted the weights on Hugging Face under the repository name moonshotai/Kimi-K3.1-Pro — no access gate, no form, no waitlist. The download is 96 safetensors shards totalling about 1.56 terabytes.
Kimi K3.1 Pro itself is not brand new. Moonshot announced it on October 23, 2026, as the first model in the K3.1 family. What is new is that the full weights are now public.
The numbers
The headline figure is 2.8 trillion total parameters, which makes K3.1 Pro the largest publicly downloadable model. It is a mixture-of-experts model: it has 896 experts, and only 16 of them activate for any given token, so 104 billion parameters are active at a time.
The context window is 1 million tokens — 1,048,576, to be exact. In 24 of the model's 93 layers, the usual attention mechanism is replaced by a different design, which is part of how it handles that long context.
The license: open for most of us, with a line drawn at the top
"Open weights" and "open source" are not the same thing. The weights ship under Moonshot's own Kimi K3.1-Pro License, not a standard open-source license like Apache 2.0. For independent developers, small teams, and most agencies, that licence is permissive enough to use and build on. Read it before you build a product on it, especially at larger scale.
The benchmarks
Moonshot's own numbers put K3.1 Pro near the top of the field. On Terminal-Bench 4.0, a coding-agent benchmark, it performed strongly. As with every vendor benchmark table, treat those numbers as a starting point, not proof — they measure what an agent can do, which is the result that matters most for developers.
What it takes to run: this is not a laptop model
The download button is the easy part. The weights are about 1.56 TB, and running them locally needs serious GPU hardware. The hosted price is genuinely competitive: $2 per million input tokens and $10 per million output tokens.
What this means for WordPress and Elementor builders
The single most useful thing about a model like K3.1 Pro for a WordPress developer is the combination of long context and strong coding ability. A 1-million-token context means a long-context code audit, or a debugging session across a whole theme or plugin codebase, is realistic in one conversation.
The open weights matter for a second reason specific to WordPress: sites that handle client data can, in principle, run capable models on infrastructure they control, rather than sending everything to a hosted API.
What I would do this week
First, do not download 1.56 terabytes on a whim. Try it hosted first — on Kimi.ai, through the API, or through a hosted provider.
Second, if you run a WordPress agency, look at the long-context coding angle before anything else. Try it on a real task: a long-context code audit, or a debugging session across a whole theme.
Third, read the license before you build a product on it. For personal use, small teams, and most agencies, the terms are workable — but check them yourself.
Kimi K3.1 Pro's weights release is the rare AI story this week where the headline and the reality line up — with fine print worth reading.
Read the full article on ConvertElementor: https://convertelementor.com/blog/moonshot-just-put-a-2-8-trillion-parameter-model-on-hugging-face-kimi-k3-1-pro-s-open-weights-are-real-and-the-fine-print-matters-november-2026-20261105
Top comments (0)