TextImage + text

MiniMax

Large MoE models with very long context (up to 1M tokens for Text-01 and M3). M3 is multimodal and understands images. In October the open weights of M3.1-Flash came out (about 428B, 23B active, up to 1M tokens of context, text, images and video): a model for programming, agents and computer use, sped up by sparse attention. MiniMax Community License: free for non-commercial use; commercial use needs the caption "Built with MiniMax M3.1-Flash" and a one-time notice to MiniMax, and a separate permission when the product earns $20 million a year or more; military use is prohibited. The benchmark figures come from MiniMax itself. Licenses differ greatly from version to version.

Developer
MiniMax, China
First release
Jan 2025
Latest release
Oct 2026
Sizes
230B-A10B – 456B-A46B
License
Commercial use with conditionsMixed: M1: Apache 2.0; M2 and M2.5: modified MIT; M2.7: non-commercial only; M3: MiniMax Community License (notify MiniMax for commercial use; from $20 million annual revenue, separate permission is required)
Russian
Supported
Ready-made builds
GGUF, AWQ, MLX (Apple)
Running
On your own serverNeeds a GPU
Industries
Legal, Software development, Finance, Media and production
Deploy on your server Hugging Face GitHub

What it does

  • Analysis of large document archives in a single request
  • Agents with tools
  • Help for developers
  • Conversational characters and scenarios

Where it is used

LawyersAnalyticsSoftware developmentContent and entertainment

Hardware requirements

LaptopLaptop or regular PC, up to 8 GB of VRAM — smaller versions
no versions
1 GPUOne GPU with 16–80 GB — mid-size versions
no versions
ClusterServer with several GPUs — flagship versions
fits

Versions

  1. MiniMax-M3.1-Flash (428B-A23B)
  2. MiniMax-M3
  3. MiniMax-M2.7
  4. MiniMax-M2.5
  5. MiniMax-M2.1
  6. MiniMax-M2
  7. MiniMax-M1
  8. MiniMax-Text-01

How to run it

On your own serverWeights are downloaded from Hugging Face and served with vLLM (load and API) or llama.cpp (modest hardware). The model then runs in a closed network with no per-request fees.Hugging Face
How much hardware you needCalculate the VRAM for the model size, context length and number of concurrent requests.Open the hardware calculator

I can set this up end to end: pick the model size, deploy it on your server and connect it to your systems. Quantization compresses a model so it takes less video memory and runs on more modest hardware. Answers change slightly, so quality is checked on your own examples.

Frequently asked questions

Can MiniMax be used in a commercial project?

With conditions. License: Mixed: M1: Apache 2.0; M2 and M2.5: modified MIT; M2.7: non-commercial only; M3: MiniMax Community License (notify MiniMax for commercial use; from $20 million annual revenue, separate permission is required). Restrictions vary — region, company revenue, attribution requirements. Have a lawyer check the terms before a commercial launch.

What hardware does MiniMax need?

At minimum: Server with several GPUs — flagship versions. Without a GPU the model is not practical. You can calculate the exact VRAM for your model size and context in the hardware calculator.

Does MiniMax support Russian?

Yes, Russian is listed on the model card.

Where can I download MiniMax and what does it cost?

The MiniMax weights are open and free to download. You only pay for the hardware it runs on and for the setup. Source links are at the bottom of this page.

How I deploy it for clients

  1. SelectionI pick the model size for your task and hardware and test it on your examples.
  2. DeploymentI deploy it on your server or in a closed network and provide an API.
  3. Fine-tuningI fine-tune it on your data (LoRA) or connect a knowledge base — whichever is cheaper for the task.
  4. IntegrationI connect it to your CRM, ERP, bot, website or team chat and set up monitoring.

Similar models

TextGLM (ChatGLM)Zhipu AI (Z.ai) · ChinaCommercial use with conditions

One of the oldest Chinese open lines: from ChatGLM-6B to GLM-5.3. Strong at agentic tasks and programming; GLM-5.3-Flash understands images and is released under MIT.

DetailsTextKimiMoonshot AI · ChinaCommercial use with conditions

Very large Moonshot MoE models for agentic work. K3 (2.8 trillion parameters) was the largest open model at release, with up to 1M tokens of context and image understanding; K2.7-Code is built for programming. On November 4 the weights of K3.1 Pro were opened (2.8 trillion parameters, 104B active, 1M tokens of context, images): a refinement of K3 for programming, 3D and financial research; by the tests of Moonshot itself, Terminal-Bench 4.0 is 65.7%. The weights are in MXFP4, about 1.56 TB, run through vLLM and SGLang, and need a serious cluster.

DetailsTextDeepSeekDeepSeek · ChinaCommercial use allowed

DeepSeek's flagship line: from the first 7B/67B to V4-Pro with 1.6 trillion parameters. Closed-model quality under an open MIT license; V4-Flash-Vision-Exp and V4.1-Flash understand images, context up to 1M tokens. The fresh V4.1-Flash-1017 answers shorter and faster than the previous Flash: by the measurements of DeepSeek itself, 77.0 vs 74.2 on DeepSWE v1.1 and 42.3 vs 31.2 on Terminal-Bench 4.0.

Details

Source: huggingface.co/MiniMaxAI/MiniMax-M3. Data checked against the model card on 22 Sep 2026. Have a lawyer review the license before commercial launch.