June 12, 2026 at 2:07 PM
MiniMax has release the MiniMax M2.7 model weights.
Blog: www.minimax.io/news/minimax...
Model: huggingface.co/MiniMaxAI/Mi...
Blog: www.minimax.io/news/minimax...
Model: huggingface.co/MiniMaxAI/Mi...
MiniMaxAI/MiniMax-M2.7 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
April 12, 2026 at 2:54 AM
MiniMax has release the MiniMax M2.7 model weights.
Blog: www.minimax.io/news/minimax...
Model: huggingface.co/MiniMaxAI/Mi...
Blog: www.minimax.io/news/minimax...
Model: huggingface.co/MiniMaxAI/Mi...
August 3, 2026 at 3:55 AM
December 26, 2025 at 3:21 PM
@astrra.space happy days huggingface.co/MiniMaxAI/Mi...
MiniMaxAI/MiniMax-M2.5 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
February 13, 2026 at 2:17 PM
@astrra.space happy days huggingface.co/MiniMaxAI/Mi...
MiniMax H3: a 30B omni-modal LLM
inputs: text, image, audio, video
outputs: text, audio, video
Note that only one model here is open weights, the H3-Context-IR component is critical yet API-only. Less critical H3-Regenerate-2K is also closed
huggingface.co/MiniMaxAI/Mi...
inputs: text, image, audio, video
outputs: text, audio, video
Note that only one model here is open weights, the H3-Context-IR component is critical yet API-only. Less critical H3-Regenerate-2K is also closed
huggingface.co/MiniMaxAI/Mi...
MiniMaxAI/MiniMax-H3 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
August 3, 2026 at 1:10 PM
MiniMax H3: a 30B omni-modal LLM
inputs: text, image, audio, video
outputs: text, audio, video
Note that only one model here is open weights, the H3-Context-IR component is critical yet API-only. Less critical H3-Regenerate-2K is also closed
huggingface.co/MiniMaxAI/Mi...
inputs: text, image, audio, video
outputs: text, audio, video
Note that only one model here is open weights, the H3-Context-IR component is critical yet API-only. Less critical H3-Regenerate-2K is also closed
huggingface.co/MiniMaxAI/Mi...
MiniMax has released the weight and you may be able to run this on a single M3 Ultra with 512Gb - the memory requirement is 460GB.
vLLM Memory requirements: 220 GB for weights, 240 GB per 1M context tokens
huggingface.co/MiniMaxAI/Mi...
vLLM Memory requirements: 220 GB for weights, 240 GB per 1M context tokens
huggingface.co/MiniMaxAI/Mi...
February 13, 2026 at 2:31 PM
MiniMax has released the weight and you may be able to run this on a single M3 Ultra with 512Gb - the memory requirement is 460GB.
vLLM Memory requirements: 220 GB for weights, 240 GB per 1M context tokens
huggingface.co/MiniMaxAI/Mi...
vLLM Memory requirements: 220 GB for weights, 240 GB per 1M context tokens
huggingface.co/MiniMaxAI/Mi...
Minimax is CRAZY
As Dean W Ball said, open source AI is deccelerationist, so obviously this is all anti-AI praxis.
huggingface.co/MiniMaxAI/Mi...
As Dean W Ball said, open source AI is deccelerationist, so obviously this is all anti-AI praxis.
huggingface.co/MiniMaxAI/Mi...
MiniMaxAI/MiniMax-Music3 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
August 13, 2026 at 5:11 PM
Minimax is CRAZY
As Dean W Ball said, open source AI is deccelerationist, so obviously this is all anti-AI praxis.
huggingface.co/MiniMaxAI/Mi...
As Dean W Ball said, open source AI is deccelerationist, so obviously this is all anti-AI praxis.
huggingface.co/MiniMaxAI/Mi...
MiniMax M2.5 is now available on @hf.co
huggingface.co/MiniMaxAI/Mi...
✨ 229B - Modified MIT license
✨37% faster than M2.1
✨ ~$1/hour at 100 TPS
huggingface.co/MiniMaxAI/Mi...
✨ 229B - Modified MIT license
✨37% faster than M2.1
✨ ~$1/hour at 100 TPS
February 13, 2026 at 2:40 PM
MiniMax M2.5 is now available on @hf.co
huggingface.co/MiniMaxAI/Mi...
✨ 229B - Modified MIT license
✨37% faster than M2.1
✨ ~$1/hour at 100 TPS
huggingface.co/MiniMaxAI/Mi...
✨ 229B - Modified MIT license
✨37% faster than M2.1
✨ ~$1/hour at 100 TPS
every six months another Chinese AI company announces they have an open source engine that more or less matches the cutting edge models and then the market forgets about it and goes back to thinking the winner of the AI race prints infinite money huggingface.co/MiniMaxAI/Mi...
December 27, 2025 at 12:55 AM
every six months another Chinese AI company announces they have an open source engine that more or less matches the cutting edge models and then the market forgets about it and goes back to thinking the winner of the AI race prints infinite money huggingface.co/MiniMaxAI/Mi...
🎵MiniMax-Music3 (open-weight)
Next-Generation Open-Weights Production-Ready & Versatile Music Model
Model: huggingface.co/MiniMaxAI/Mi...
Repo: github.com/MiniMax-AI/M...
Next-Generation Open-Weights Production-Ready & Versatile Music Model
Model: huggingface.co/MiniMaxAI/Mi...
Repo: github.com/MiniMax-AI/M...
August 13, 2026 at 4:17 PM
🎵MiniMax-Music3 (open-weight)
Next-Generation Open-Weights Production-Ready & Versatile Music Model
Model: huggingface.co/MiniMaxAI/Mi...
Repo: github.com/MiniMax-AI/M...
Next-Generation Open-Weights Production-Ready & Versatile Music Model
Model: huggingface.co/MiniMaxAI/Mi...
Repo: github.com/MiniMax-AI/M...
MiniMax-M3 just dropped 🔥
huggingface.co/MiniMaxAI/Mi...
✨ 428B / 23B active
✨ 1M context
✨ MiniMax Sparse Attention (MSA)
And it’s not just weights!
- paper: huggingface.co/papers/2606....
- kernel: huggingface.co/kernels/Mini...
- Transformers support
Love how this was released❤️
huggingface.co/MiniMaxAI/Mi...
✨ 428B / 23B active
✨ 1M context
✨ MiniMax Sparse Attention (MSA)
And it’s not just weights!
- paper: huggingface.co/papers/2606....
- kernel: huggingface.co/kernels/Mini...
- Transformers support
Love how this was released❤️
June 12, 2026 at 2:40 PM
MiniMax-M3 just dropped 🔥
huggingface.co/MiniMaxAI/Mi...
✨ 428B / 23B active
✨ 1M context
✨ MiniMax Sparse Attention (MSA)
And it’s not just weights!
- paper: huggingface.co/papers/2606....
- kernel: huggingface.co/kernels/Mini...
- Transformers support
Love how this was released❤️
huggingface.co/MiniMaxAI/Mi...
✨ 428B / 23B active
✨ 1M context
✨ MiniMax Sparse Attention (MSA)
And it’s not just weights!
- paper: huggingface.co/papers/2606....
- kernel: huggingface.co/kernels/Mini...
- Transformers support
Love how this was released❤️
fwiw, this is explicitly considered by the license huggingface.co/MiniMaxAI/Mi.... i think this is what a healthy open ecosystem will look like, actually.
August 27, 2026 at 5:22 PM
fwiw, this is explicitly considered by the license huggingface.co/MiniMaxAI/Mi.... i think this is what a healthy open ecosystem will look like, actually.
The audio is weird speech-like garbage, I think that's because I didn't prompt it to tell it what audio to add - there are prompting instructions here which I didn't find until after I'd made the video huggingface.co/MiniMaxAI/Mi...
August 4, 2026 at 7:17 PM
The audio is weird speech-like garbage, I think that's because I didn't prompt it to tell it what audio to add - there are prompting instructions here which I didn't find until after I'd made the video huggingface.co/MiniMaxAI/Mi...
MiniMax-M1 🔥 The first reasoning model by MiniMax AI
is now live on @hf.co
huggingface.co/collections/...
✨ 40k/80k thinking budget
✨ Powered by Hybrid MoE + Lightning Attention 👀
✨ 1M context length 🤯
✨ Apache 2.0
✨ RL-trained for math, coding & real-world software
is now live on @hf.co
huggingface.co/collections/...
✨ 40k/80k thinking budget
✨ Powered by Hybrid MoE + Lightning Attention 👀
✨ 1M context length 🤯
✨ Apache 2.0
✨ RL-trained for math, coding & real-world software
MiniMax-M1 - a MiniMaxAI Collection
MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model.
huggingface.co
June 16, 2025 at 3:30 PM
MiniMax-M1 🔥 The first reasoning model by MiniMax AI
is now live on @hf.co
huggingface.co/collections/...
✨ 40k/80k thinking budget
✨ Powered by Hybrid MoE + Lightning Attention 👀
✨ 1M context length 🤯
✨ Apache 2.0
✨ RL-trained for math, coding & real-world software
is now live on @hf.co
huggingface.co/collections/...
✨ 40k/80k thinking budget
✨ Powered by Hybrid MoE + Lightning Attention 👀
✨ 1M context length 🤯
✨ Apache 2.0
✨ RL-trained for math, coding & real-world software
Ah, covered in my newsletter but the weights came out after :) Minimax H3 model plus spaces for trying it:
MiniMaxAI/MiniMax-H3 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
August 3, 2026 at 6:05 AM
Ah, covered in my newsletter but the weights came out after :) Minimax H3 model plus spaces for trying it: