Why We Built VIBE Bench: Rethinking Evaluation for Real Workloads
β’
4
AI & ML interests
Intelligence with Everyone
Recent Activity
View all activity
Articles
Organization Card
Welcome to the official Hugging Face organization for MiniMax.
Together, we are building artificial general intelligence with everyone, ensuring AI is accessible to everyone, and addressing the challenges of tomorrow responsibly.
Here are our tech report and paper:
- MiniMax M2.1 Tech Blogs:
- VTP: Towards Scalable Pre-training of Visual Tokenizers for Generation
- MiniMax M2 Tech Blogs:
- Tech Blog 1: What makes good reasoning data
- Tech Blog 2: Aligning to What? Rethinking Agent Generalization in MiniMax M2
- Tech Blog 3: Why Did MiniMax M2 End Up as a Full Attention Model?
- MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention
- SynLogic: Synthesizing Verifiable Reasoning Data at Scale for Learning Logical Reasoning and Beyond
- One RL to See Them All: Visual Triple Unified Reinforcement Learning
- MiniMax-Speech: Intrinsic Zero-Shot Text-to-Speech with a Learnable Speaker Encoder
- MiniMax-01: Scaling Foundation Models with Lightning Attention
Explore our latest innovations and experience:
Towards Scalable Pre-training of Visual Tokenizers for Generation
-
MiniMaxAI/VTP-Small-f16d64
Image Feature Extraction β’ 0.2B β’ Updated β’ 16.7k β’ 11 -
MiniMaxAI/VTP-Base-f16d64
Image Feature Extraction β’ 0.3B β’ Updated β’ 15.6k β’ 18 -
MiniMaxAI/VTP-Large-f16d64
Image Feature Extraction β’ 0.7B β’ Updated β’ 18.7k β’ 13 -
Towards Scalable Pre-training of Visual Tokenizers for Generation
Paper β’ 2512.13687 β’ Published β’ 100
MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model.
-
MiniMax M1
π¬350Generate code from text prompts
-
MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention
Paper β’ 2506.13585 β’ Published β’ 273 -
MiniMaxAI/MiniMax-M1-80k
Text Generation β’ 456B β’ Updated β’ 218 β’ β’ 687 -
MiniMaxAI/MiniMax-M1-40k
Text Generation β’ 456B β’ Updated β’ 13k β’ 182
Towards Scalable Pre-training of Visual Tokenizers for Generation
-
MiniMaxAI/VTP-Small-f16d64
Image Feature Extraction β’ 0.2B β’ Updated β’ 16.7k β’ 11 -
MiniMaxAI/VTP-Base-f16d64
Image Feature Extraction β’ 0.3B β’ Updated β’ 15.6k β’ 18 -
MiniMaxAI/VTP-Large-f16d64
Image Feature Extraction β’ 0.7B β’ Updated β’ 18.7k β’ 13 -
Towards Scalable Pre-training of Visual Tokenizers for Generation
Paper β’ 2512.13687 β’ Published β’ 100
MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model.
-
MiniMax M1
π¬350Generate code from text prompts
-
MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention
Paper β’ 2506.13585 β’ Published β’ 273 -
MiniMaxAI/MiniMax-M1-80k
Text Generation β’ 456B β’ Updated β’ 218 β’ β’ 687 -
MiniMaxAI/MiniMax-M1-40k
Text Generation β’ 456B β’ Updated β’ 13k β’ 182
spaces
5
Running
Featured
350
MiniMax M1
π¬
Generate code from text prompts
Running
103
MiniMax Speech Tech Report
π
Generate high-quality speech from text with voice cloning
Running
64
MiniMaxVL01
π¬
Generate responses to text and images in a chat interface
Running
118
MiniMaxText01
π¬
Generate responses to text and images in a chat interface
models
15
MiniMaxAI/MiniMax-M2.1
Text Generation
β’
229B
β’
Updated
β’
199k
β’
β’
945
MiniMaxAI/MiniMax-M2
Text Generation
β’
229B
β’
Updated
β’
123k
β’
β’
1.44k
MiniMaxAI/VTP-Large-f16d64
Image Feature Extraction
β’
0.7B
β’
Updated
β’
18.7k
β’
13
MiniMaxAI/VTP-Base-f16d64
Image Feature Extraction
β’
0.3B
β’
Updated
β’
15.6k
β’
18
MiniMaxAI/VTP-Small-f16d64
Image Feature Extraction
β’
0.2B
β’
Updated
β’
16.7k
β’
11
MiniMaxAI/MiniMax-M1-40k-hf
Text Generation
β’
456B
β’
Updated
β’
82
β’
11
MiniMaxAI/MiniMax-Text-01-hf
Text Generation
β’
456B
β’
Updated
β’
8.77k
β’
9
MiniMaxAI/MiniMax-M1-80k-hf
Text Generation
β’
456B
β’
Updated
β’
67
β’
7
MiniMaxAI/MiniMax-M1-80k
Text Generation
β’
456B
β’
Updated
β’
218
β’
β’
687
MiniMaxAI/MiniMax-M1-40k
Text Generation
β’
456B
β’
Updated
β’
13k
β’
182