Kimi K3.1 Pro

Description

Kimi K3.1 Pro is a 3T-class multimodal model from Moonshot AI for reasoning, coding, and agentic workflows. It accepts text, images, and video and generates text, with a 1,048,576-token context window.

This model is ready for commercial or non-commercial use.

Third-Party Community Consideration:

This model is not owned or developed by NVIDIA. This model has been developed and built to a third-party's requirements for this application and use case; see link to Non-NVIDIA Kimi K3.1 Pro Model Card

License and Terms of Use:

GOVERNING TERMS: The trial service is governed by the NVIDIA API Trial Terms of Service; and the use of this model is governed by the NVIDIA Open Model Agreement. ADDITIONAL INFORMATION: The base model Kimi K3.1 Pro is governed by the Kimi K3.1 Pro License.

Deployment Geography:

Global

Use Case:

Use Case: Developers can use the model for code generation, image and video understanding, retrieval-augmented generation, document question answering, and tool-assisted reasoning. Other documented applications include CAD scripting and financial research.

Release Date:

Build.NVIDIA.com: 11/04/2026 via link

Reference(s):

References:

• Kimi K3.1 Pro announcement and evaluation methodology
• Kimi K3.1 Pro developer documentation
• Kimi model list
• Kimi image and video input documentation

Model Architecture:

Architecture Type: Transformer
Network Architecture: Mixture-of-Experts
Total Parameters: 2.8 × 10^12 (2.8T)
Active Parameters: 1.04 × 10^11 (104B)
Vocabulary Size: 160,000
Base Model: Kimi K3

Input:

Input Types: Text, Image, Other (Video)
Input Formats: String, Red, Green, Blue (RGB), Other (Video: MP4/WebM)
Input Parameters: One-Dimensional (1D), Two-Dimensional (2D), Three-Dimensional (3D)
Other Input Properties: Moonshot recommends images up to 4096 × 2160 pixels and video up to 1920 × 1080 pixels; these are processing recommendations, not hard limits. Input Context Length (ISL): 1048576

Output:

Output Types: Text
Output Format: String
Output Parameters: One-Dimensional (1D)
Other Output Properties: The model context window is 1,048,576 tokens; a separate maximum output length is undisclosed.

Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.

Software Integration:

Runtime Engines:

• Other: vLLM with NVIDIA Dynamo 1.7.0

Supported Hardware Microarchitecture Compatibility:

• NVIDIA Blackwell: GB300

Supported Operating System(s): Linux

The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.

Model Version(s)

Kimi K3.1 Pro (v3.1). Vendor model identifier: kimi-k3.1-pro.

Training, Testing, and Evaluation Datasets:

Training Dataset

Data Modality: Text, Image, Video
Image Training Data Size: Undisclosed
Text Training Data Size: Undisclosed
Video Training Data Size: Undisclosed
Training Data Collection: Undisclosed
Training Labeling: Undisclosed
Training Properties: Post-training uses joint reinforcement learning across hundreds of domains. The developer reports removing cyber-related mid-training data and reinforcement-learning tasks; dataset names and modality-specific sizes are undisclosed.

Testing Dataset

Testing Data Collection: Undisclosed
Testing Labeling: Undisclosed
Testing Properties: Undisclosed

Evaluation Dataset

Evaluation Data Collection: Hybrid: Automated, Manually-Collected
Evaluation Labeling: Hybrid: Automated, Manually-Labeled
Evaluation Properties: Moonshot AI reports evaluations of coding, CAD reconstruction, and financial-document tasks.
Evaluation Benchmark Score: Vendor-reported results are listed below; scores use the associated agent and tool configurations.

Evaluation Methodology Notes:

• Benchmark testing conditions: Reasoning effort max, temperature 1.0, and top-p 0.95, unless otherwise specified.
• Terminal-Bench 4.0: Claude-Code harness; resource-constrained terminal tasks with strict task-specific tests.
• BenchCAD: Multi-view CAD reconstruction on a 1,000-file subset; Voxel IoU scores distinguish tool-assisted and unassisted runs.
• OfficeQA Pro v2: Claude Code; the entire PDF corpus is rendered as images without machine-readable text. Tasks that retrieve reference answers online receive zero.

See the vendor announcement for the complete benchmark suite and methodology.

Inference

Acceleration Engine: vLLM
Test Hardware: NVIDIA Blackwell (GB300 × 8)

Additional Details

Cybersecurity

Moonshot AI reports reduced vulnerability-exploitation capability after training-data filtering, while residual dual-use cybersecurity capability remains.

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. Developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

Please make sure you have proper rights and permissions for all input image and video content; if image or video includes people, personal health information, or intellectual property, the image or video generated will not blur or maintain proportions of image subjects included.

Users are responsible for model inputs and outputs. Users are responsible for ensuring safe integration of this model, including implementing guardrails as well as other safety mechanisms, prior to deployment.

Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns here