--- title: MiniMax-M3-MXFP8 canonical_url: "https://www.modelscope.ai/models/MiniMax/MiniMax-M3-MXFP8" md_url: "https://www.modelscope.ai/models/MiniMax/MiniMax-M3-MXFP8.md" repository: MiniMax/MiniMax-M3-MXFP8 last_updated: 2026-07-12 license: other pipeline_tag: image-text-to-text tasks: - image-text-to-text model_type: - minimax_m3_vl architectures: - MiniMaxM3SparseForConditionalGeneration base_model: - MiniMaxAI/MiniMax-M3 base_model_relation: quantized parameters: 440.3B tensor_type: - U8 - F8_E4M3 - BF16 - F32 library_name: - safetensors - pytorch frameworks: - pytorch downloads: 115 stars: 5 tags: - multimodal - moe - agent - coding - video --- # MiniMax-M3-MXFP8 > MiniMax-M3-MXFP8 - An open-source model by MiniMax on ModelScope. MiniMax-M3 is a native multimodal model with 1M context. It has 428B parameters and 23B activated parameters. MiniMax/MiniMax-M3-MXFP8 is a 440.3B-parameter image-text-to-text model on ModelScope. licensed under other. derived from MiniMaxAI/MiniMax-M3. - **Repository**: MiniMax/MiniMax-M3-MXFP8 - **License**: other - **Tasks**: image-text-to-text - **Parameters**: 440.3B - **Base model**: MiniMaxAI/MiniMax-M3 - **Tags**: multimodal, moe, agent, coding, video - **Downloads**: 115 - **Stars**: 5 - **Last updated**: 2026-07-12 Source: https://www.modelscope.ai/models/MiniMax/MiniMax-M3-MXFP8 ---
MiniMax

MiniMax Agent API MiniMax Website
ModelScope MiniMax AI WeChat Discord Hugging Face GitHub arXiv Paper LICENSE

MiniMax-M3 is a native multimodal model with 1M context. It has ~428B parameters and ~23B activated parameters. **Highlights:** - **Native Multimodality:** M3 undergoes mixed-modality training from the very first step, enabling deeper semantic fusion across text, image, and video. - **Context Scaling via Sparse Attention:** M3 introduces MiniMax Sparse Attention (MSA) to improve long context efficiency. M3 delivers 9ร— prefill and 15ร— decode speedups compared to M2 at 1M context, reducing per-token compute to 1/20. - **Coding & Cowork Capability:** M3 achieves frontier-level performance across long-horizon agentic benchmarks, excelling in both coding and cowork. MiniMax-M3-MXFP8 is the MXFP8 quantized variant of [MiniMax-M3](https://huggingface.co/MiniMaxAI/MiniMax-M3), a native multimodal model with 1M context. It has ~428B parameters and ~23B activated parameters.

## MiniMax Sparse Attention (MSA) M3 is powered by [**MiniMax Sparse Attention (MSA)**](https://github.com/MiniMax-AI/MSA), a high-performance sparse attention operator designed for million-token contexts. Compared with GQA, MSA dramatically reduces the attention compute and memory footprint while preserving model quality.

GQA vs MSA Efficiency Comparison

> ๐Ÿ“„ Read the technical report: [arXiv:2606.13392](https://arxiv.org/abs/2606.13392) ยท [Hugging Face Papers](https://huggingface.co/papers/2606.13392) ## How to Use - [MiniMax Agent](https://agent.minimax.io/) - [MiniMax API](https://platform.minimax.io/) M3 supports three reasoning modes through the `thinking` parameter: - **`enabled`** โ€” Reasoning is always enabled. - **`adaptive`** โ€” M3 automatically determines when additional reasoning is beneficial. - **`disabled`** โ€” Reasoning is disabled to minimize latency and maximize throughput. ## Local Deployment Download the model: ```bash hf download MiniMaxAI/MiniMax-M3 --local-dir MiniMax-M3 ``` We recommend the following inference frameworks to serve the model: - [SGLang](https://docs.sglang.io/) - see [SGLang cookbook](https://docs.sglang.io/cookbook/autoregressive/MiniMax/MiniMax-M3). - [vLLM](https://github.com/vllm-project/vllm) - see [vLLM recipes](https://recipes.vllm.ai/MiniMaxAI/MiniMax-M3). - [Transformers](https://github.com/huggingface/transformers) - see [Transformers docs](https://huggingface.co/docs/transformers/model_doc/minimax_m3_vl). ### Inference Parameters We recommend the following parameters for best performance: `temperature=1.0`, `top_p=0.95`. ## Contact Us Contact us at [model@minimax.io](mailto:model@minimax.io).