Use MiniMax\-H3 directly via App\. - WebApp Global: hailuoai\.video \| CN: hailuoai\.com - Desktop Global: hub\.minimax\.io \| CN: hub\.minimaxi\.com MiniMax H3 is a general-purpose, omni-modal generative system. It supports unified understanding of multimodal contexts composed of text, images, video, and audio, and can generate video with native stereo audio at resolutions up to 2K and durations of up to 15 seconds. Thanks to its task-generalization-oriented system design, H3 already possesses broad multimodal context understanding and generation capabilities at the pre-training stage, enabling outstanding performance in following complex multimodal instructions. The complete H3 system consists of the following three modules: - H3-Context-IR: As inputs become increasingly complex, we build a dedicated system to deeply understand and refine the input multimodal instructions, then convert them into a form that H3 can readily understand—the Context Intermediate Representation—for generation. **H3-Context-IR is critical to the quality of the final output, so we strongly recommend incorporating it into your generation pipeline or following the “Prompting Guidance” to build your own context-processing system.** - H3-Base: Generates audio and video based on the H3-Context-IR output, producing results at 768p resolution. - H3-Regenerate-2K: Feeds the 768p result together with the original context back into H3 to regenerate the output at 2K resolution. This process leverages both H3’s powerful generative capabilities and the rich information contained in the original context, enabling it to produce high-resolution outputs with more accurate details and greater visual fidelity.
We introduce MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model. MiniMax-M1 is powered by a hybrid Mixture-of-Experts (MoE) architecture combined with a lightning attention mechanism. The model is developed based on our previous MiniMax-Text-01 model, which contains a total of 456 billion parameters with 45.9 billion parameters activated per token. Consistent with MiniMax-Text-01, the M1 model natively supports a context length of 1 million tokens, 8x the context size of DeepSeek R1. Furthermore, the lightning attention mechanism in MiniMax-M1 enables efficient scaling of test-time compute – For example, compared to DeepSeek R1, M1 consumes 25% of the FLOPs at a generation length of 100K tokens. These properties make M1 particularly suitable for complex tasks that require processing long inputs and thinking extensively. MiniMax-M1 is trained using large-scale reinforcement learning (RL) on diverse problems ranging from traditional mathematical reasoning to sandbox-based, real-world software engineering environments. We develop an efficient RL scaling framework for M1 highlighting two perspectives: (1) We propose CISPO, a novel algorithm that clips importance sampling weights instead of token updates, which outperforms other competitive RL variants; (2) Our hybrid-attention design naturally enhances the efficiency of RL, where we address unique challenges when scaling RL with the hybrid architecture. We train two versions of MiniMax-M1 models with 40K and 80K thinking budgets respectively. Experiments on standard benchmarks show that our models outperform other strong open-weight models such as the original DeepSeek-R1 and Qwen3-235B, particularly on complex software engineering, tool using, and long context tasks. With efficient scaling of test-time compute, MiniMax-M1 serves as a strong foundation for next-generation language model agents to reason and tackle real-world challenges. Benchmark performance comparison of leading commercial and open-weight models across competition-level mathematics, coding, software engineering, agentic tool use, and long-context understanding tasks. We use the MiniMax-M1-80k model here for Mini
Extensively trained with reinforcement learning in hundreds of thousands of complex real-world environments, M2.5 is **SOTA in coding, agentic tool use and search, office work, and a range of other economically valuable tasks**, boasting scores of **80.2% in SWE-Bench Verified, 51.3% in Multi-SWE-Bench, and 76.3% in BrowseComp** (with context management). Trained to reason efficiently and decompose tasks optimally, M2.5 exhibits tremendous speed in performing complicated agentic tasks, completing the SWE-Bench Verified evaluation **37% faster** than M2.1, matching the speed of **Claude Opus 4.6**. M2.5 is the first frontier model where users do not need to worry about cost, delivering on the promise of intelligence too cheap to meter. **It costs just $1 to run the model continuously for an hour at a rate of 100 tokens per second.** At 50 tokens per second, the cost drops to $0.30. We hope that the speed and cost effectiveness of M2.5 enable innovative new agentic applications.
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 9× prefill and 15× 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. M3 is powered by **MiniMax Sparse Attention (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.
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 9× prefill and 15× 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, a native multimodal model with 1M context. It has ~428B parameters and ~23B activated parameters.
**MiniMax Music 3** is a high-performance music generation model for creating complete songs up to **five minutes** long. Conditioned on lyrics and a detailed music description, it generates structurally coherent songs with expressive vocals, evolving arrangements, and stable long-form audio quality. MiniMax Music 3 combines an **8B Global LLM** for long-range musical structure, a **0.6B Local LLM** for frame-level acoustic detail, and a continuous hidden-state synthesis system based on **Flow Matching** and **Flow-VAE**. The model produces 32 kHz, 16-bit stereo WAV audio. MiniMax Music 3 natively supports full-song generation up to five minutes. The model maintains musical themes, rhythm, vocal identity, and arrangement progression across long sequences, enabling complete structures such as intro, verse, pre-chorus, chorus, bridge, instrumental break, and outro.
This repository is primarily for the Transformers framework. If you're using other open-source frameworks, please use the alternative repository: MiniMax-Text-01 MiniMax-Text-01 is a powerful language model with 456 billion total parameters, of which 45.9 billion are activated per token. To better unlock the long context capabilities of the model, MiniMax-Text-01 adopts a hybrid architecture that combines Lightning Attention, Softmax Attention and Mixture-of-Experts (MoE). Leveraging advanced parallel strategies and innovative compute-communication overlap methods—such as Linear Attention Sequence Parallelism Plus (LASP+), varlen ring attention, Expert Tensor Parallel (ETP), etc., MiniMax-Text-01's training context length is extended to 1 million tokens, and it can handle a context of up to 4 million tokens during the inference. On various academic benchmarks, MiniMax-Text-01 also demonstrates the performance of a top-tier model. The architecture of MiniMax-Text-01 is briefly described as follows: - Total Parameters: 456B - Activated Parameters per Token: 45.9B - Number Layers: 80 - Hybrid Attention: a softmax attention is positioned after every 7 lightning attention. - Number of attention heads: 64 - Attention head dimension: 128 - Mixture of Experts: - Number of experts: 32 - Expert hidden dimension: 9216 - Top-2 routing strategy - Positional Encoding: Rotary Position Embedding (RoPE) applied to half of the attention head dimension with a base frequency of 10,000,000 - Hidden Size: 6144 - Vocab Size: 200,064
We are delighted to introduce our **MiniMax-VL-01** model. It adopts the "ViT-MLP-LLM" framework, which is a commonly used technique in the field of multimodal large language models. The model is initialized and trained with three key parts: a 303-million-parameter Vision Transformer (ViT) for visual encoding, a randomly initialized two-layer MLP projector for image adaptation, and the MiniMax-Text-01 as the base LLM. MiniMax-VL-01 has a notable dynamic resolution feature. Input images are resized per a pre-set grid, with resolutions from 336×336 to 2016×2016, keeping a 336×336 thumbnail. The resized images are split into non-overlapping patches of the same size. These patches and the thumbnail are encoded separately and then combined for a full image representation. The training data for MiniMax-VL-01 consists of caption, description, and instruction data. The Vision Transformer (ViT) is trained on 694 million image-caption pairs from scratch. Across four distinct stages of the training pipeline, a total of 512 billion tokens are processed, leveraging this vast amount of data to endow the model with strong capabilities. Finally, MiniMax-VL-01 has reached top-level performance on multimodal leaderboards, demonstrating its edge and dependability in complex multimodal tasks. For production deployment, we recommend using vLLM to serve MiniMax-VL-01. vLLM provides excellent performance for serving large language models with the following features: 🔥 Outstanding service throughput performance ⚡ Efficient and intelligent memory management 📦 Powerful batch request processing capability ⚙️ Deeply optimized underlying performance For detailed deployment instructions, please refer to our vLLM Deployment Guide. MiniMax-VL-01 supports Function Calling capability, enabling the model to intelligently identify when external functions need to be called and output parameters in structured JSON format. With Function Calling, you can:
