llmrun
lyber-aiKimi K3

Kimi K3.1 Pro UNCENSORED MXFP4 — Hardware Requirements & GPU Compatibility

ChatReasoning

Kimi K3.1 Pro UNCENSORED MXFP4 is a 2780.1B-parameter open language model from lyber-ai in the Kimi K3 family. At Q4_K_M it needs about 1834.87 GB of VRAM — see which GPUs and Macs can run it below.

2780.1B params · denseOther13 downloads33 likes
Based on Kimi K3.1 Pro

Kimi K3.1 Pro UNCENSORED MXFP4 runs on a GPU cluster

Q4_K_M needs ~1.8 TB · about 4 nodes of 8× H100 SXM 80 GB

Other

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Kimi K3.1 Pro UNCENSORED MXFP4

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 GPU nodes
tensor + pipeline parallel · needs vLLM on every node
$vllm serve lyber-ai/Kimi-K3.1-Pro-UNCENSORED-MXFP4 --max-model-len 16384 --tensor-parallel-size 8 --pipeline-parallel-size <nodes>

Size the nodes from the table above

Download: Weights on Hugging Face

Ray cluster setup, other ways

Ways to runOther ways

each opens step-by-step

Run in cloud

Rent 4× 8-GPU H100 SXM nodes · Q4_K_M

Estimated hourly price
~$86.08/hr
est., whole cluster
GPUs
32 GPUs
2.6 TB of VRAM
Deployon RunPodRent H100 SXM$2.69/hr per GPU (opens in new tab)

$5+ bonus for new users

Compare GPUs →

How Much VRAM Does Kimi K3.1 Pro UNCENSORED MXFP4 Need?

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

QuantizationVRAM+32KRuns on
1299.7 GB1328 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsRuns on 8× B200 192 GB
1490.8 GB1519.1 GB3-bit medium — 3-bit medium quantizationRuns on 3 nodes of 8× H100 SXM
1834.9 GB1863.2 GB4-bit medium — 4-bit medium quantization — most popular sweet spotRuns on 4 nodes of 8× H100 SXM
2178.9 GB2207.2 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffRuns on 4 nodes of 8× H100 SXM
2522.9 GB2551.3 GB6-bit, very good — 6-bit quantization, very good qualityRuns on 5 nodes of 8× H100 SXM
3058.1 GB3086.4 GB8-bit, near-lossless — 8-bit quantization, near-losslessRuns on 6 nodes of 8× H100 SXM
6116.2 GB6144.5 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingRuns on 11 nodes of 8× H100 SXM
est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

1848.1GBat 16K context

Weights
1832.7 GB
KV cache
15.1 GB
Overhead
0.3 GB

Server-class: ~1.8 TB at 16K context · about 4 nodes of 8× H100 SXM 80 GB

Which Hardware Can Run Kimi K3.1 Pro UNCENSORED MXFP4?

Q4_K_M · 1834.9 GB

Kimi K3.1 Pro UNCENSORED MXFP4 (Q4_K_M) requires 1834.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 2386+ GB is recommended. No consumer GPU has enough memory.

Needs server-class hardware

Kimi K3.1 Pro UNCENSORED MXFP4 runs on a GPU cluster · Q4_K_M needs ~1.8 TB

Hardware that runs this model
HardwareQuantMemoryContextHardware page
8× B200 192 GB · 1.5 TBQ2_K1312.9 GB32K ctxB200 →Q2_K · 1312.9 GB · 32K ctx
GPU cluster · 4 nodes of 8× H100 SXM 80 GBQ4_K_M1848.1 GB32K ctxH100 SXM →Q4_K_M · 1848.1 GB · 32K ctx

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

≤ 20736 GB tier →

By VRAM tier · typical GPUs and Macs at Q4_K_M

Consumer GPUs (up to 32 GB) and Macs up to 512 GB: too small for this model (needs ≥ 1.3 TB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →

Alternatives to Kimi K3.1 Pro UNCENSORED MXFP4

Kimi K3 family · 3 models →

Specifications

Publisher
lyber-ai
Family
Kimi K3
Parameters
2780.1B
Release Date
2026-11-06
License
Other

Resources

Frequently Asked Questions

How much VRAM does Kimi K3.1 Pro UNCENSORED MXFP4 need?

Kimi K3.1 Pro UNCENSORED MXFP4 requires 1834.9 GB of VRAM at Q4_K_M, or 6116.2 GB at BF16.

VRAM = Weights + KV Cache + Overhead

Weights = 2780.1B × 4.8 bits ÷ 8 = 1668.1 GB

KV Cache + Overhead ≈ 166.8 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.

VRAM usage by quantization

1834.9 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run Kimi K3.1 Pro UNCENSORED MXFP4?

No — Kimi K3.1 Pro UNCENSORED MXFP4 requires at least 1299.7 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

What's the best quantization for Kimi K3.1 Pro UNCENSORED MXFP4?

For Kimi K3.1 Pro UNCENSORED MXFP4, Q4_K_M (1834.9 GB) offers the best balance of quality and VRAM usage. Q5_K_M (2178.9 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 1299.7 GB.

VRAM requirement by quantization

Q2_K
1299.7 GB
Q4_K_M ★
1834.9 GB
Q5_K_M
2178.9 GB
Q6_K
2522.9 GB
Q8_0
3058.1 GB
BF16
6116.2 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Kimi K3.1 Pro UNCENSORED MXFP4 on a Mac?

Kimi K3.1 Pro UNCENSORED MXFP4 requires at least 1299.7 GB at Q2_K. None of the 159 Macs we list has that much usable memory (the largest has 508.5 GB after macOS overhead).

Can I run Kimi K3.1 Pro UNCENSORED MXFP4 locally?

Yes — Kimi K3.1 Pro UNCENSORED MXFP4 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

What's the download size of Kimi K3.1 Pro UNCENSORED MXFP4?

At Q4_K_M, the download is about 1668.07 GB. The full-precision BF16 version is 5560.22 GB. The smallest option (Q2_K) is 1181.55 GB.

Which GPUs can run Kimi K3.1 Pro UNCENSORED MXFP4?

No single consumer GPU has enough VRAM to run Kimi K3.1 Pro UNCENSORED MXFP4 at Q4_K_M (1834.9 GB). Multi-GPU or professional hardware is required.

Which devices can run Kimi K3.1 Pro UNCENSORED MXFP4?

Kimi K3.1 Pro UNCENSORED MXFP4 requires at least 1834.9 GB at Q4_K_M, which exceeds the usable memory of every unified-memory device we list (the largest has 508.5 GB).