Kimi K3.1 Pro
Moonshot AI · China · #2 in our ranking · Most capable
Kimi K3.1 Pro is an open-weight mixture-of-experts model from Moonshot AI (China), released Nov 2026, with 2.8T parameters (104B active per token), a 1M-token context window, and the Kimi K3.1 Pro License.
Kimi K3.1 Pro is a post-trained version of Kimi K3, released with open weights on November 2, 2026. It keeps the 2.8T-parameter mixture-of-experts design with 104B active parameters and a 1M-token context, and swaps the full-attention layers for sparse attention so long contexts run more efficiently. It reads text and images, and Moonshot trained it for coding, financial research, and other long agent tasks.
Kimi K3.1 Pro specs
- Provider
- Moonshot AI
- Origin
- China
- Architecture
- Mixture-of-experts
- Parameters
- 2.8T total · 104B active
- Context
- 1M tokens | 786K words
- License
- Kimi K3.1 Pro License
- Input
- Text, Image
- Released
- Nov 2026
- Weights
- Hugging Face
Sourced from the model card on Hugging Face. Last checked 2026-11-11.
Highlights
- 104B active parameters
- Top open-weight Terminal-Bench 4.0 score in Moonshot's report
- 1M-token context
Where it fits for payment teams
- Chargebacks and disputes
#2 pick. The most capable open model in our list, reading text and images with a 1M-token context, for complex, document-heavy cases.
Harnesses in Runtime
Run Kimi K3.1 Pro through any of these harnesses on Runtime.
Provider-reported benchmarks
| Terminal-Bench 4.0 | 65.7 |
|---|---|
| DeepSWE v1.1 | 75.5 |
| Automation Bench-AA | 73.3 |
Benchmark scores are reported by each model provider on its Hugging Face model card. They are not independently verified by Runtime.
How to run Kimi K3.1 Pro
Download the weights
Pull moonshotai/Kimi-K3.1-Pro from Hugging Face. Check the Kimi K3.1 Pro License.
Serve it in your cloud
vLLM or SGLang on your GPUs, or a managed host.
Put it to work in Runtime
Point an agent or skill at it, then compare with evals.
Kimi K3.1 Pro FAQ
What is the context window of Kimi K3.1 Pro?
+
Kimi K3.1 Pro supports a 1M-token context window (1,048,576 tokens), roughly 786K words of English text.
How many parameters does Kimi K3.1 Pro have?
+
Kimi K3.1 Pro is a mixture-of-experts model with 2.8T total parameters, of which 104B are active per token.
Is Kimi K3.1 Pro free for commercial use?
+
Kimi K3.1 Pro is released under the Kimi K3.1 Pro License, a custom license. Commercial use may come with conditions, so read the license on Hugging Face before you deploy or fine-tune it.
What can Kimi K3.1 Pro take as input?
+
Kimi K3.1 Pro accepts text, image input and generates text.
Is Kimi K3.1 Pro good for chargebacks and disputes?
+
Yes. Kimi K3.1 Pro is our #2 pick for chargebacks and disputes. The most capable open model in our list, reading text and images with a 1M-token context, for complex, document-heavy cases.
Can I run Kimi K3.1 Pro in my own cloud with Runtime?
+
Yes. Download the weights from Hugging Face (moonshotai/Kimi-K3.1-Pro) and serve them with vLLM or SGLang in your cloud, or use a managed provider that hosts the model. Runtime agents can then use Kimi K3.1 Pro through a harness like OpenCode, with your data staying in your environment.