- Context
- 1Mtokens
- Released
- 1y ago
- Size
- 456Bparameters
MiniMax-M1 is an open-weight reasoning model from MiniMax, built on the company's earlier MiniMax-Text-01 model and introduced as the first large-scale open-source hybrid-attention reasoning model. MiniMax states that for deep reasoning the model needs only about 30% of the computing power of DeepSeek R1, which it credits to its hybrid lightning-attention design. The model targets complex productivity tasks such as software engineering, long-context understanding, and agentic tool use, and is positioned as a foundation for reasoning agents.
- Language model
- Open weights · 456B
- 1M context
- Reasoning
- Tools
How it compares
Additional source evidence
- Source model
- minimax/minimax-m1-80k
- Source family
- minimax
- Last updated
- 2025-06-17
- Model context
- 1M
- Model output limit
- 80K
- Open weights
- Yes
- File attachments
- No
- Reasoning
- Yes
- Tool calling
- Yes
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "minimax/minimax-m1-80k",
"name": "MiniMax M1 80K",
"description": "Open hybrid-attention MoE reasoning model with a million-token context and an 80K thinking budget",
"family": "minimax",
"last_updated": "2025-06-17",
"attachment": false,
"reasoning": true,
"tool_call": true,
"temperature": true,
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"limit": {
"context": 1000000,
"output": 80000
},
"release_date": "2025-06-17",
"open_weights": true,
"license": "Apache-2.0",
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/MiniMaxAI/MiniMax-M1-80k"
}
]
}Frequently asked questions
Running locally
MiniMax M1 80K needs an estimated 318.5 GB of VRAM at Q4, 557.8 GB at Q8 and 1,049.5 GB at FP16. These figures are estimates from the 456B parameter count, covering weights plus about 15% and 0.5 GB for a short context.
No. Estimates show MiniMax M1 80K does not fit on a 24 GB, 16 GB, 12 GB or 8 GB GPU at Q4, Q8 or FP16.
Yes, on very large Macs. Assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is an estimated 512 GB, since Q4 needs an estimated 318.5 GB of VRAM.
Privacy and data
Yes. MiniMax M1 80K has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window of MiniMax M1 80K is 1M tokens (1,000,000) and its maximum output is 80K tokens (80,000).
No. MiniMax M1 80K only accepts text as input.
Yes. MiniMax M1 80K supports tool calling (function calling).
Yes. MiniMax M1 80K supports a reasoning mode.
Quality and benchmarks
Closest alternatives, matched on model kind and release date (no comparable price), are ERNIE 4.5 0.3B by Baidu, ERNIE 4.5 VL 28B A3B by Baidu and DiffuCoder 7B by Apple.
ERNIE 4.5 0.3BMiniMax M1 80K vs ERNIE 4.5 0.3BERNIE 4.5 VL 28B A3BMiniMax M1 80K vs ERNIE 4.5 VL 28B A3BDiffuCoder 7BMiniMax M1 80K vs DiffuCoder 7B
Licensing and access
MiniMax M1 80K has open weights, released under the Apache-2.0 license.
Yes. MiniMax M1 80K is licensed under Apache 2.0, a permissive license that allows commercial use.
The weights are on Hugging Face in the repository MiniMaxAI/MiniMax-M1-80k.
Basics
MiniMax made MiniMax M1 80K, and it was released on June 17, 2025.
Yes. The listed facts state that MiniMax M1 80K has open weights.
Generated from Vertical Tabs data, updated November 4, 2026.
Discussion
No comments yet
Be the first to share your thoughts.