back to local model checker

MiniMax M3

MoEToolsThinkingVisionPopular

MiniMax · 428B (23B active) · Mixture of Experts

Native multimodal MoE — understands text, image and long video with 1M context

MiniMax Community
Downloads

0

Likes

0

Released

2026-06

Context

1M tokens

Use cases

chatvisionreasoningcode
Total experts128
Active experts4
Active params23.0B

Your hardware

16 GB Reference Desktop

Quant
Q4_K_M
Memory
219.7 GB
Speed
--
Fit
Too heavy

Quantization options

What is quantization?
QuantBitsVRAMDiskQualityStatus
Q2_K2137.5 GB130.8 GBlowToo heavy · ~1 t/s
Q3_K_M3192.3 GB183.1 GBmoderateToo heavy · ~1 t/s
Q4_K_M4219.7 GB209.3 GBgoodToo heavy · ~1 t/s
Q5_K_M5274.5 GB261.6 GBgoodToo heavy · ~1 t/s
Q6_K6329.4 GB313.9 GBexcellentToo heavy · ~1 t/s
Q8_08439 GB418.5 GBexcellentToo heavy · ~1 t/s
F1616877.4 GB837.1 GBlosslessToo heavy · ~1 t/s

About this model

MiniMax M3 (MiniMax m3) is a 428B sparse mixture-of-experts (MoE) model activating 23B active per token from MiniMax in the MiniMax family, designed primarily for chat, vision, reasoning, code workflows. Native reasoning/thinking tokens expand output length per query, so reserve extra KV cache headroom. Supports structured tool calling; use Q4_K_M or higher when strict JSON schema adherence matters. Supports up to 1M context tokens; start at 4K–8K with Q8_0 KV cache on tight GPUs.

Hardware planning baselines start at 239.2 GB minimum memory and 398.6 GB recommended memory. At Q4_K_M, weights plus baseline runtime overhead use roughly 219.7 GB of RAM/VRAM and 209.3 GB of disk space. At 428B, MiniMax M3 belongs firmly in the workstation and multi-GPU tier. Running it comfortably requires 398.6 GB of total memory across dual RTX GPUs, high-memory Apple Silicon (Mac Studio Ultra/Max), or MoE hybrid CPU-RAM offloading—mobile devices and standard 16 GB laptops are too constrained for practical speeds.

Deploy MiniMax M3 locally via llama.cpp with Flash Attention enabled. Under MiniMax Community, review official weight cards before commercial redistribution.

Highlights

  • Local Fit Baseline239.2 GB minimum and 398.6 GB recommended RAM/VRAM.
  • Primary QuantQ4_K_M targets 219.7 GB memory and 209.3 GB disk.
  • Context & Capabilities1M max context · Sparse MoE · Tool calling · Reasoning.

System requirements by device

Memory planning targets 239.2 GB minimum RAM/VRAM and 398.6 GB recommended memory. The baseline quant is Q4_K_M, estimated at 219.7 GB memory and 209.3 GB disk. Actual runtime usage shifts with context length, KV cache precision, and OS background overhead.

DeviceNotes
Mac (Apple silicon)Requires a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra with enough unified wired budget for 219.7 GB.
Windows / Linux GPUUse dual/multi-GPU CUDA splitting (--tensor-split) or MoE CPU expert offloading (-ot) paired with 32–128 GB DDR5 system RAM.
Phones & Compact Edge BoardsNot practical on phones (239.2 GB min RAM exceeds Android LMK and iOS Jetsam limits). Pick a 2B–4B model on mobile.

Start with a short session and leave memory for the operating system. A file fitting on disk does not guarantee that it will load in memory.

Estimated memory by quantization

These are rough estimates, not tested minimum requirements. Check the actual file and runtime documentation; context, offloading and operating-system overhead can change the memory needed.

Quant TierEstimated memoryDisk SizeQuality
Q2_K137.5 GB130.8 GBlow
Q3_K_M192.3 GB183.1 GBmoderate
Q4_K_M219.7 GB209.3 GBgood
Q5_K_M274.5 GB261.6 GBgood
Q6_K329.4 GB313.9 GBexcellent
Q8_0439.0 GB418.5 GBexcellent
F16877.4 GB837.1 GBlossless

How to run MiniMax M3 on your device

Start with a short prompt and the smallest quant that still gives the quality you need. A clean first run is more useful than loading a larger file with no memory headroom.

Windows

Run MiniMax M3 on Multi-GPU or High-RAM Desktop

  1. Plan for 239.2 GB minimum combined memory and 398.6 GB recommended memory for Q4_K_M.
  2. Use llama.cpp `-ot ".ffn_.*_exps.=CPU"` or `--tensor-split` across multiple GPUs so attention and KV cache remain in fast VRAM.
  3. Enable Flash Attention (`-fa on`) and Q8_0 KV cache (`-ctk q8_0 -ctv q8_0`) to prevent context buffers from exceeding GPU headroom.
  4. Avoid attempting to run workstation-scale weights on smartphones or 8 GB laptops; use a 2B–8B checkpoint on portable devices.
Device how-to guide
Mac

Run MiniMax M3 on High-Memory Apple Silicon Mac

  1. Verify you are on a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra capable of holding 219.7 GB in wired unified memory.
  2. Launch via `llama-server` or LM Studio with MLX/Metal enabled and set initial context conservatively to 4K–8K.
  3. Monitor Activity Monitor Memory Pressure to confirm zero SSD swap thrashing during token generation.
  4. Step down to a lighter quantization (e.g., Q3_K_M or IQ4_XS) if total memory utilization exceeds 75%.
Device how-to guide

MiniMax M3 vs alternatives

ModelParamsContextMin RAMLicense
MiniMax M3428B (23B active)1M239.2 GBMiniMax Community
MiniMax H333B32K18.4 GBMiniMax Community
Llama 3.1 405B405B128K226.3 GBLlama 3.1 Community
Llama 4 Maverick 17B-128E400B (17B active)1M223.5 GBLlama 4 Community
Qwen 3.5 397B-A17B397B (17B active)256K221.8 GBApache 2.0
Qwen 3 Coder 480B480B (35B active)256K268.2 GBApache 2.0

Use this table to choose the smallest nearby model that still matches the job. A related model with lower RAM can be the better local choice when the current page fits only at the edge of available memory.

Frequently asked questions

What is MiniMax M3?
MiniMax M3 is a 428B sparse mixture-of-experts model (23B active active per token) from MiniMax, supporting up to 1M tokens with 239.2 GB minimum RAM and 398.6 GB recommended RAM.
How do I download and install MiniMax m3?
Download a GGUF build for MiniMax M3, start with Q4_K_M, and confirm the model fits locally before increasing context or quant quality.
How much VRAM or RAM does MiniMax M3 require?
At Q4_K_M, MiniMax M3 occupies roughly 219.7 GB of active memory and 209.3 GB on disk. Add 2–6 GB of spare headroom for KV cache buffers, runtime scratch space, and OS apps (398.6 GB recommended).
What is the best way to run MiniMax M3 locally?
Run MiniMax M3 in llama.cpp starting with Q4_K_M. Keep initial context at 4K–8K tokens and confirm 100% GPU/Metal offload before scaling context length.
Can MiniMax M3 run on a consumer laptop or smartphone?
No smartphone or 8–16 GB laptop can hold MiniMax M3 cleanly in local memory. It targets multi-GPU NVIDIA workstations, high-RAM Mac Studios, or MoE CPU-RAM hybrid rigs with at least 239.2 GB total memory.
How does MiniMax M3 compare to MiniMax H3?
MiniMax H3 (33B, 18.4 GB min RAM) sits closest in size. Compare active parameters, context ceiling (1M vs 32K), and license terms (MiniMax Community) before switching.