---
format: "aidr-story-markdown/v1"
id: "eae8dac5afc1915f48dc3386278161340f157edcf636adba3a3817a50b8041df"
canonical_url: "https://aidr.today/eae8dac5?lang=en"
title: "MiniMax Opens Weights for 428B Parameter M3.1 Flash Model"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-23T01:41:02.000Z"
category: "Models"
topics: ["llm","open-source","devtools"]
source_urls: ["https://huggingnews.com/ai/minimax-opens-weights-for-428b-parameter-m31-flash-model-684c3e3f","https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash","https://x.com/MiniMax_AI/status/2113436622889029887","https://x.com/MiniMax_AI/status/2112150867868541221","https://x.com/MaxForAI/status/2112160282495914376","https://x.com/kimmonismus/status/2112151168579227892","https://x.com/teortaxesTex/status/2112152366200021378","https://x.com/ValsAI/status/2112403307075698859"]
summary: "MiniMax published the underlying weights of its 428 billion parameter M3.1 Flash model on October 23 through Hugging Face and ModelScope, ending a stealth phase as the \"Space Bunny\" prototype that had processed more than 200 trillion tokens. The model supports a 1 million token context window and is specialized for multi-step automation, coding and cybersecurity, achieving inference speeds exceeding 1,000 tokens per second on Nvidia Blackwell and AMD Instinct GPUs using the SGLang Omni framework."
---

# MiniMax Opens Weights for 428B Parameter M3\.1 Flash Model

> [Open the canonical story](<https://aidr.today/eae8dac5?lang=en>)

**Published:** 2026-10-23T01:41:02.000Z
**Category:** Models
**Topics:** llm, open\-source, devtools

## Summary

MiniMax published the underlying weights of its 428 billion parameter M3\.1 Flash model on October 23 through Hugging Face and ModelScope, ending a stealth phase as the "Space Bunny" prototype that had processed more than 200 trillion tokens\. The model supports a 1 million token context window and is specialized for multi\-step automation, coding and cybersecurity, achieving inference speeds exceeding 1,000 tokens per second on Nvidia Blackwell and AMD Instinct GPUs using the SGLang Omni framework\.

## Sources

- [Story source](<https://huggingnews.com/ai/minimax-opens-weights-for-428b-parameter-m31-flash-model-684c3e3f>)
- [Story source](<https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash>)
- [Story source](<https://x.com/MiniMax_AI/status/2113436622889029887>)
- [Story source](<https://x.com/MiniMax_AI/status/2112150867868541221>)
- [Supporting source](<https://x.com/MaxForAI/status/2112160282495914376>)
- [Supporting source](<https://x.com/kimmonismus/status/2112151168579227892>)
- [Supporting source](<https://x.com/teortaxesTex/status/2112152366200021378>)
- [Supporting source](<https://x.com/ValsAI/status/2112403307075698859>)