README.md

---
pipeline_tag: text-generation
base_model:
- MiniMaxAI/MiniMax-M3
license: other
license_name: nvidia-open-model-license
license_link: https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license
library_name: Model Optimizer
tags:
- nvidia
- ModelOpt
- MiniMax-M3
- DSpark
---
# Model Overview

## Description:
The NVIDIA MiniMax-M3 DSpark model is the DSpark draft head of MiniMax's MiniMax-M3 model, which is a native multimodal mixture-of-experts language model that uses an optimized transformer architecture. For more information, please check [here](https://huggingface.co/MiniMaxAI/MiniMax-M3). The NVIDIA MiniMax-M3 DSpark model incorporates DSpark speculative decoding with [Model Optimizer](https://github.com/NVIDIA/Model-Optimizer).

This model is ready for commercial/non-commercial use.  <br>

## License/Terms of Use:
Governing Terms: Use of this model is governed by the [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).

**ADDITIONAL INFORMATION** : [MiniMax Community License](https://huggingface.co/MiniMaxAI/MiniMax-M3/blob/main/LICENSE). **MiniMax-M3** .

## Deployment Geography:
Global

## Use Case:
Developers and inference providers building multimodal AI agents, coding assistants, long-context reasoning systems, and cowork-style assistants that benefit from lower-latency speculative decoding. This model is intended for DSpark-assisted serving of MiniMax-M3 for text, image, and video understanding workflows.

## Release Date:
Hugging Face 07/22/2026 via [https://huggingface.co/nvidia/MiniMax-M3-DSpark](https://huggingface.co/nvidia/MiniMax-M3-DSpark)

## References

- Nvidia Model Optimizer: [https://github.com/NVIDIA/Model-Optimizer](https://github.com/NVIDIA/Model-Optimizer)
- [MiniMax-M3 base model card](https://huggingface.co/MiniMaxAI/MiniMax-M3)
- [MiniMax Sparse Attention technical report](https://huggingface.co/papers/2606.13392)
- [DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation](https://huggingface.co/papers/2607.05147)

## Model Architecture:
**Architecture Type:** Transformers  <br>
**Network Architecture:** MiniMaxM3SparseForConditionalGeneration <br>
**Number of Model Parameters:** 428B in total and 23B activated <br>

<!--
## Computational Load
**Cumulative Compute:** 1.2*10^16 FLOPS
**Estimated Energy and Emissions for Model Training:** 0.31 tCO2e
-->

## Input:
**Input Type(s):** Text <br>
**Input Format(s):** String <br>
**Input Parameters:** One-Dimensional (1D), Two-Dimensional (2D), Three-Dimensional (3D) <br>
**Other Properties Related to Input:** Context length up to 1,048,576 tokens<br>

## Output:
**Output Type(s):** Text <br>
**Output Format:** String <br>
**Output Parameters:** One Dimensional(1D): Sequences <br>
**Other Properties Related to Output:** Outputs may include natural-language responses, code, structured reasoning, tool-use content, and multimodal analysis depending on application-level tooling. <br>


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:
**Supported Runtime Engine(s):** <br>
* vLLM <br>

**Supported Hardware Microarchitecture Compatibility:** <br>
* NVIDIA Blackwell <br>

**Preferred Operating System(s):** <br>
* Linux <br>

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):
**The model is DSpark version and is trained with nvidia-modelopt **v0.45.0**  <br>

## Training and Evaluation Datasets:

## Training Dataset:
**Link**: [Nemotron-Post-Training-Dataset-v2](https://huggingface.co/datasets/nvidia/Nemotron-Post-Training-Dataset-v2), only prompts from the datasets were used for data synthesis, (the original responses from GPT were not used), which is then used to train the DSpark modules. <br>
**Data Modality**:  Text, Image, Video <br>
**Image Training Data Size:** Undisclosed <br>
**Text Training Data Size:** [1 Billion to 10 Trillion Tokens] <br>
**Video Training Data Size:** Undisclosed <br>
**Data Collection Method by dataset**: Hybrid: Automated, Synthetic<br>
**Labeling Method by dataset**: Hybrid: Automated, Synthetic<br>
**Properties:** 2M multilingual text samples featuring prompts spanning math, code, STEM, and conversational topics. Each sample includes a synthetic response generated by the target model.

## Evaluation Dataset:
**Link**: MTBench, for more details, see [here](https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge); [SPEED-Bench](https://huggingface.co/datasets/nvidia/SPEED-Bench) <br>
**Data Collection Method by dataset**: Hybrid: Human, Synthetic<br>
**Labeling Method by dataset**:Hybrid: Human, Synthetic <br>
**Properties:** MT-Bench contains 3,300 multi-turn dialogue sequences, each annotated with expert preference votes. SPEED-Bench is a unified, diverse benchmark for speculative decoding spanning coding, humanities, math, multilingual, QA, RAG, reasoning, roleplay, STEM, summarization, and writing domains.

## Inference:
**Acceleration Engine:** vLLM <br>
**Test Hardware:** NVIDIA B300 <br>


## DSpark Speculative Decoding
Synthesized data was obtained from MiniMax's MiniMax-M3 model, which is then used to finetune the DSpark modules. This model is ready for inference with vLLM in DSpark speculative decoding mode, where a parallel draft backbone proposes a block of candidate tokens, a lightweight Markov head injects intra-block token dependency, and a confidence head predicts per-position acceptance probability for scheduled verification. Compared with a purely parallel drafter, DSpark is designed to improve accepted length while preserving the latency benefits of block drafting. The longest accepted candidate sequence is selected so that more than 1 token is returned in the generation step. The number of tokens generated in each step is called acceptance rate.

## Usage

To serve the checkpoint with [vLLM](https://github.com/vllm-project/vllm):

```sh
vllm serve MiniMaxAI/MiniMax-M3 \
  --tensor-parallel-size 4 \
  --block-size 128 \
  --trust-remote-code \
  --speculative-config '{
    "method": "dspark",
    "model": "<draft-model>",
    "num_speculative_tokens":8
  }'
```

Alternatively, with the Python `LLM` API:

```python
from vllm import LLM, SamplingParams

llm = LLM(
    model="MiniMaxAI/MiniMax-M3",
    tensor_parallel_size=4,
    block_size=128,
    trust_remote_code=True,
    speculative_config={
        "method": "dspark",
        "model": "<draft-model>",
        "num_speculative_tokens":8
    },
)
```

Note: `--block-size 128` (the KV-cache block size) is required by MiniMax-M3's sparse attention.

### Training

This checkpoint was trained in two phases: short-context training (4k sequence length) on M3-synthesized Speculative-Decoding-Dataset-v2, followed by a long-context continuation (16k sequence length) on a 12k+-token subset of the same corpus, which improves long-context acceptance (throughput-32k) while preserving short-context quality.

See the following streaming recipe for [NVIDIA Model Optimizer](https://github.com/NVIDIA/Model-Optimizer) as a reference for training your own DSpark draft head:

```sh
uv run launch.py --yaml examples/MiniMaxAI/MiniMax-M3/hf_streaming_dspark_multi_node.yaml --yes
```

Run from [`tools/launcher`](https://github.com/NVIDIA/Model-Optimizer/tree/main/tools/launcher); see its README for setup.

## Evaluation

Acceptance rate on [SPEED-Bench](https://huggingface.co/datasets/nvidia/SPEED-Bench) (qualitative subset) with a draft block size of 8:

| Category        | SPEED-Bench Acceptance Rate |
|-----------------|:-----------------------:|
| coding          |      **5.26**                      |
| humanities      |      **3.54**                      |
| math            |      **4.91**                      |
| multilingual    |      **5.04**                      |
| qa              |      **3.15**                      |
| rag             |      **4.75**                      |
| reasoning       |      **4.21**                      |
| roleplay        |      **2.22**                      |
| stem            |      **3.98**                      |
| summarization   |      **3.83**                      |
| writing         |      **2.84**                      |
| **Overall Average** |   **3.97**                     |

Acceptance rate on [SPEED-Bench](https://huggingface.co/datasets/nvidia/SPEED-Bench) (throughput-32k subset, long-context) with a draft block size of 8:

| Category      | SPEED-Bench Acceptance Rate |
|---------------|:-----------------------:|
| low_entropy   |      **3.19**                      |
| mixed         |      **3.74**                      |
| high_entropy  |      **2.89**                      |
| **Overall Average** |   **3.27**                     |

Acceptance rate on [MT-bench](https://huggingface.co/spaces/lmsys/mt-bench) with a draft block size of 8:

| Category    | MT Bench Acceptance Rate |
|-------------|:-----------------------:|
| writing     |           **3.22**                   |
| roleplay    |           **2.56**                   |
| reasoning   |           **3.93**                   |
| math        |           **5.78**                   |
| coding      |           **5.08**                   |
| extraction  |           **5.45**                   |
| stem        |           **3.21**                   |
| humanities  |           **2.63**                   |
| **Overall Average** |   **3.98**                   |

## Model Limitations:

The base model was trained on data that contains toxic language and societal biases originally crawled from the internet. Therefore, the model may amplify those biases and return toxic responses especially when prompted with toxic prompts. The model may generate answers that may be inaccurate, omit key information, or include irrelevant or redundant text producing socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive.

## 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.

For more detailed information on ethical considerations for this model, please see the Model Card++ Bias, Explainability, Safety & Security, and Privacy Subcards.

Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).  <br>

SUBCARDS:

# **Explainability**

|Field:|Response:|  
|:---:|:---:|  
|Intended Task/Domain:| Text generation, code generation, reasoning, and long-horizon software engineering tasks. |  
|Model Type: |Text and Image-to-text transformer |  
|Intended Users:|This model is intended for developers, researchers, and customers building/utilizing LLMs, while balancing accuracy and efficiency.|  
|Output:|Text String(s)|  
|Describe how the model works:|Generates text by predicting the next word or token based on the context provided in the input sequence using multiple self-attention layers|  
|Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of:  |  Not Applicable|
|Technical Limitations & Mitigation:| The model was trained on data that contains toxic language and societal biases originally crawled from the internet. Therefore, the model may amplify those biases and return toxic responses especially when prompted with toxic prompts. Therefore, before deploying any applications of this model, developers should perform safety testing and tuning tailored to their specific applications of the model.|  
|Verified to have met prescribed quality standards?|Yes|  
|Performance Metrics:|Accuracy, Throughput, and user-side throughput|  
|Potential Known Risk| The model may generate answers that may be inaccurate, omit key information, or include irrelevant or redundant text producing socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive. |
|Licensing:| Your usage is governed by the following [Governing Terms: Use of this model is governed by the [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).

**ADDITIONAL INFORMATION** : [MiniMax Community License](https://huggingface.co/MiniMaxAI/MiniMax-M3/blob/main/LICENSE). **MiniMax-M3** . |

# **Bias**

|Field:|Response:|  
|:---:|:---:|  
|Participation considerations from adversely impacted groups [protected classes](https://www.senate.ca.gov/content/protected-classes) in model design and testing:|None|  
|Measures taken to mitigate against unwanted bias:|None|
|Bias Metric:|None|

# **Safety & Security**

|Field:|Response:|  
|:---:|:---:|  
|Model Application(s):|Chat, Instruction Following, Chatbot Development, Code Generation, Reasoning|  
|Describe life critical application (if present):|Not Applicable |  
|Use Case Restrictions:|Abide by the [Governing Terms: Use of this model is governed by the [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).

**ADDITIONAL INFORMATION** : [MiniMax Community License](https://huggingface.co/MiniMaxAI/MiniMax-M3/blob/main/LICENSE). **MiniMax-M3** . |  
|Model and Dataset Restrictions:|The Principle of least privilege (PoLP) is applied limiting access for dataset generation.  Restrictions enforce dataset access during training, and dataset license constraints adhered to. Model checkpoints are made available on Hugging Face, and may become available on cloud providers' model catalog.|

# **Privacy**

|Field:|Response:|  
|:---:|:---:|  
|Generatable or Reverse engineerable personal data?|No|  
|Personal data used to create this model?|No|
|Was consent obtained for any personal data used?|Not Applicable|  
|How often is dataset reviewed?|Before Release|  
|Was data from user interactions with the AI model (e.g. user input and prompts) used to train the model? | No |
|Is there provenance for all datasets used in training?|Yes|  
|Does data labeling (annotation, metadata) comply with privacy laws?|Yes| 
|Is data compliant with data subject requests for data correction or removal, if such a request was made? |  Not Applicable| 
|Applicable NVIDIA Privacy Policy|https://www.nvidia.com/en-us/about-nvidia/privacy-policy/|