MiniMax-M3

About

What MiniMax-M3 is

Model Overview Description MiniMax-M3 is a multimodal model with frontier-level coding and agentic capabilities, built on a Mixture-of-Experts architecture with a 1M-token context window. The model processes text, image, video, and computer use inputs and produces text outputs, with emphasis on long-horizon coding tasks, agentic and tool-use workflows, and long-form video understanding. The NVIDIA MiniMax-M3 NVFP4 model is quantized with Model Optimizer. This model is ready for 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 MiniMax-M3 Model Card. License/Terms of Use: GOVERNING TERMS: Use of the checkpoints is governed by the MiniMax Community License.<br Additional Information: Built with MiniMax M3. Deployment Geography: Global Use Case: Use Case: MiniMax-M3 is intended for multimodal understanding across text, image, and video; long-form video understanding (up to 30 minutes); long-horizon coding tasks (8+ hours); agentic and tool-use workflows; and design and creative tasks. The model supports two reasoning modes switchable per request: thinking mode for complex reasoning and agentic tasks, and non-thinking mode for latency-sensitive scenarios. Release Date: Hugging Face 06/23/2026 via https://huggingface.co/nvidia/MiniMax-M3-NVFP4 Model Architecture: Architecture Type: Transformer<br Network Architecture: Mixture-of-Experts (multimodal)<br Total Parameters: 428B<br Active Parameters: Approximately 23B per token (A23B)<br Vision Encoder: ViT for image and video input Input: Input Types: Text, Image, Video<br Input Formats: Text: String; Image: RGB images; Video: encoded video file<br Input Parameters: One-Dimensional (1D), Two-Dimensional (2D), Three-Dimensional (3D)<br Other Input Properties: Supports long-form video input up to 30 minutes.<br Input Context Length (ISL): 1 million tokens Output: Output Types: Text<br Output Format: String<br Output Parameters: One-Dimensional (1D)<br Other Output Properties: None Our AI models are designed and/or optimized to run on NVIDIA

Summarised from the published model card. Read the full card on the HuggingFace links below.

Specifications

What it is

How it works

How language models work

Variants

Sizes & precisions

Open weights ship in several sizes and precisions. One page, all the variants — pick the one that fits your GPU. VRAM figures are estimates from model size.

Using it via the API

Call it like any OpenAI endpoint

Once AxForge deploys nvidia-minimax-m3 for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (nvidia-minimax-m3 below is illustrative; you get the exact model name on deployment.)

$ curl -sS https://api.axforge.ai/v1/chat/completions \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"nvidia-minimax-m3","messages":[{"role":"user","content":"Hello"}]}'

Details

Languages, data & research

Licence

Commercial licence needed

Sources

Weights & code

Explore

More language models