llmrun
MiniMaxMiniMaxMiniMaxVL01ForConditionalGeneration

MiniMax VL 01 — Hardware Requirements & GPU Compatibility

VisionFunctions

MiniMax-VL-01 is MiniMax's vision-language model with roughly 456 billion total and 49 billion active parameters. It follows a ViT-MLP-LLM design that pairs a 303-million-parameter Vision Transformer and a two-layer MLP projector with the MiniMax-Text-01 language model as its base. Images are handled at dynamic resolutions from 336x336 to 2016x2016 with a thumbnail, and the card lists function calling support. The card reports training on 512 billion tokens across four stages. Local use is impractical for most setups, since it needs a multi-GPU server or a very large unified-memory machine even when quantized. The model is released under the MiniMax Model Agreement, a custom license, with the code under MIT, so the model terms should be read before commercial use. It was published in January 2025 and uses MiniMax-Text-01 as its language backbone.

456.4B params · MoE49B active8K context5.3K downloads287 likes

MiniMax VL 01 runs on a multi-GPU server

BF16 needs ~915.2 GB · about 8× H200 SXM 141 GB

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run MiniMax VL 01

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Recommended

vLLM on a multi-GPU server
tensor + pipeline parallel · needs vLLM on every node
$vllm serve MiniMaxAI/MiniMax-VL-01 --max-model-len 8192 --tensor-parallel-size 8

Size the nodes from the table above

Download: Weights on Hugging Face

install, other ways

Ways to runOther ways

each opens step-by-step

Run in cloud

Rent 8× H200 SXM node · BF16

Estimated hourly price
~$28.72/hr
est., whole server
GPUs
8 GPUs
1.1 TB of VRAM

How Much VRAM Does MiniMax VL 01 Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KRuns on
913.7 GB—16-bit · bfloat — Brain floating point 16 — preferred for trainingRuns on 8× H200 SXM 141 GB
est.calculated, no published file yet

Selected BF16

Brain floating point 16

915.2GBat 8K context

Weights
912.9 GB
KV cache
2.0 GB
Overhead
0.3 GB

Server-class: ~915.2 GB at 8K context · about 8× H200 SXM 141 GB

Which Hardware Can Run MiniMax VL 01?

BF16 · 913.7 GB

MiniMax VL 01 (BF16) requires 913.7 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 1188+ GB is recommended. Using the full 8K context window can add up to 1.5 GB, bringing total usage to 915.2 GB. No consumer GPU has enough memory.

Needs server-class hardware

MiniMax VL 01 runs on a multi-GPU server · BF16 needs ~915.2 GB

Hardware that runs this model
HardwareQuantMemoryContextHardware page
8× H200 SXM 141 GB · 1.1 TBBF16915.2 GB8K ctxH200 SXM →BF16 · 915.2 GB · 8K ctx
6× B200 192 GB · 1.2 TBBF16915.2 GB8K ctxB200 →BF16 · 915.2 GB · 8K ctx

Also: AMD Instinct MI355X Platform (8x MI355X) · NVIDIA DGX B200 · NVIDIA DGX B300 · NVIDIA DGX H200 · NVIDIA GB200 NVL72 · NVIDIA GB300 NVL72

≤ 20736 GB tier →

By VRAM tier · typical GPUs and Macs at BF16

Consumer GPUs (up to 32 GB) and Macs up to 512 GB: too small for this model (needs ≥ 915.2 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →

Alternatives to MiniMax VL 01

MiniMax family · 10 models →

Specifications

Publisher
MiniMax
Family
MiniMax
Parameters
456.4B
Active parameters
48.8B per token
Architecture
MiniMaxVL01ForConditionalGeneration
Context Length
8,192 tokens
Vocabulary Size
200,064
Release Date
2025-01-12

Frequently Asked Questions

How much VRAM does MiniMax VL 01 need?

MiniMax VL 01 requires 913.7 GB of VRAM at BF16. Full 8K context adds up to 1.5 GB (915.2 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 456.4B × 16 bits ÷ 8 = 912.9 GB

KV Cache + Overhead ≈ 0.8 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 2.3 GB (at full 8K context)

VRAM usage by quantization

913.7 GB
915.2 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run MiniMax VL 01?

No — MiniMax VL 01 requires at least 913.7 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

Can I run MiniMax VL 01 on a Mac?

MiniMax VL 01 requires at least 913.7 GB at BF16. None of the 159 Macs we list has that much usable memory (the largest has 508.5 GB after macOS overhead).

Can I run MiniMax VL 01 locally?

Yes — MiniMax VL 01 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

What's the download size of MiniMax VL 01?

At BF16, the download is about 912.87 GB.

Which GPUs can run MiniMax VL 01?

No single consumer GPU has enough VRAM to run MiniMax VL 01 at BF16 (913.7 GB). Multi-GPU or professional hardware is required.

Which devices can run MiniMax VL 01?

MiniMax VL 01 requires at least 913.7 GB at BF16, which exceeds the usable memory of every unified-memory device we list (the largest has 508.5 GB).