Unsloth 发布 MiniMax-M3.1-Flash 量化 GGUF，2-bit 版可在 150GB 内存本地运行
TL;DR · AI 摘要
Unsloth åå¸ MiniMax-M3.1-Flash éå GGUFï¼2-bit çå¯å¨ 150GB å åæ¬å°è¿è¡ · AIHOT Unsloth AI · @ Unslot...
核心要点
- 主题聚焦：Unsloth 发布 MiniMax-M3.1-Flash 量化 GGUF，2-bit 版可在
- 来源：AI HOT 精选，建议结合原文判断细节。
- AI 分析暂不可用，本条为保底评分与摘要。
Unsloth åå¸ MiniMax-M3.1-Flash éå GGUFï¼2-bit çå¯å¨ 150GB å åæ¬å°è¿è¡ · AIHOT
Unsloth AI
· @
UnslothAI
· X
·
2026-10-26 22:45
2 å°æ¶å
ç²¾é
AI è¯å
70
AI 导读
Unsloth åå¸ MiniMax-M3.1-Flash çéå GGUFï¼2-bit çå¨ç¼©å° 45% ä½ç§¯çåæ¶ä¿ç 91.5% top 1 精度ï¼å¯å¨ 150GB å åä¸æ¬å°è¿è¡ï¼1-bit ç约 89.9%ï¼4-bit UD-Q4_K_XL è¾¾ 98.0%ãä½è ç§°å ¶ä¸ºå尺寸æ强å¼æºæ¨¡åä¹ä¸ï¼å¹¶ç»åº Hugging Face æéé¾æ¥ä¸æ¬å°è¿è¡æåã
æ¨èçç±
åæç»åºèªå»º GGUF éåå¨ä¿ç精度ä¸å缩ä½ç§¯é´çå®æµæ°æ®ï¼ä»¥ååæ¡£ä½æéå åï¼è¯»è å¯æ®æ¤å¤æè½å¦æ¬å°é¨ç½²ã
æ£æ · AI ç¿»è¯
MiniMax-M3.1-Flash ç°å¨å¯ä»¥å¨æ¬å°è¿è¡äºï¼
æ们让 2-bit GGUF å¨ä½ç§¯ç¼©å° 45% çåæ¶ï¼ä»ä¿ç 91.5% ç top 1 åç¡®çãå¯å¨ 150GB å åä¸è¿è¡ã
MiniMax-M3.1-Flash æ¯å尺寸ä¸æ强çå¼æºæ¨¡åï¼æ´ä½ä¸ä¹å±äºæä½³ä¹åã
GGUFï¼ https://huggingface.co/unsloth/MiniMax-M3.1-Flash-GGUF æåï¼ https://unsloth.ai/docs/models/minimax-m3.1-flash
å¼ç¨ MiniMax (official) @ MiniMax_AI
MiniMax M3.1 Flashï¼å¤ªç©ºçº§æ§è½ + å å级é度 ðð° 没éï¼Space Bunnyââæ¾ä»¥ 200T+ tokens ç»é¡¶å ¨çæè¡æ¦ââæ£æ¯ M3.1 Flash çæ©æçæ¬ã ð°éæ°å®ä¹æºè½ä½ AI çæ§è½ç»æµå¦ãM3.1 Flash å¨ Coworkï¼Automation-Benchï¼62%ï¼ALEï¼30%ï¼ãç¼ç¨ï¼DeepSWEï¼74%ï¼åè§è§æ¨çï¼BabyVisionï¼91%ï¼çå ³é®ä»»å¡ä¸è¾¾å°å沿级æºè½æ°´å¹³ï¼ææ¬ä» 为 M3 ç 1/3ï¼é度æå 3 åã API å®ä»·ï¼ç»ä¸è´¹çï¼ï¼æ¯ 1M è¾å ¥/è¾åº tokens åå«ä¸º $0.15 / $0.60ï¼ç¼å读å为 $0.015 / 1M tokensã 模åæéå°äº 10 æ 23 æ¥åå¸ã APIï¼http://platform.minimax.io M Planï¼https://www.minimax.io/m-plan MiniMax Code æ¡é¢çï¼https://agent.minimax.io/download
åæ
MiniMax M3.1 Flash: Space Performance + Bunny Speed ðð° Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash. ð°Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed. API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens. Model weights will arrive on Oct 23. API: http://platform.minimax.io M Plan: https://www.minimax.io/m-plan MiniMax Code for desktop: https://agent.minimax.io/download
å¨ X æ¥ç被å¼ç¨çå¸å
æ¥æºï¼ Unsloth AI · x.com
MiniMax
端侧 AI
å¼æºçæ
产åæ´æ°
#
端侧