AMD Instinct MI350X

288 GB VRAM2,300 FP16 TFLOPS1,000 W TDP0 providers0 offerings0 regions

The AMD Instinct MI350X is an enterprise-class GPU built on the CDNA 4 architecture, featuring 288 GB of memory with 8,000 GB/s of memory bandwidth and a 1,000 W TDP. It delivers 144.2 TFLOPS of FP32 performance and 2,300 TFLOPS of FP16 tensor performance.

Cheapest

—

No current offerings

Median

—

across 0 offerings

Most expensive

—

90-day trend

-42.9%

median rate

Hardware specifications

Same across all providers

Memory

288 GB

What LLMs fit? ↓

FP16 performance

2,300 TFLOPS

peak

TDP

1000 W

Memory bandwidth

8,000 GB/s

Architecture

CDNA 4

FP32 performance

144.2 TFLOPS

Brand

AMD

Series

Instinct MI

Market price history

Median price across all providers

Prices updated 5 hours ago

0 offerings from 0 providers

Sorted by price ascending. Compare configurations side-by-side.

Provider	Count	vCPU	RAM	Region	Per GPU hour	Total/hr	Action
Try broadening your filters or clear all filters to start over.

LLMs that fit on the AMD Instinct MI350X

Estimated VRAM for popular open-weight LLMs at 16-bit (FP16/BF16), 8-bit (FP8/INT8) and 4-bit (INT4/MXFP4) precision, against 288 GB per GPU. Showing the 12 largest that fit.

Model	Parameters	16-bit	8-bit	4-bit
DeepSeek V4 Flash DeepSeek	284B	— Not released	— Not released	192 GB 1× · 96 GB free
Solar Open2 250B Upstage	250B 15B active	601 GB Too large	330 GB Tight on 1× · 4% spare	192 GB 1× · 96 GB free
Qwen3 235B-A22B Alibaba	235B 22B active	564 GB Too large	310 GB Tight on 1× · 11% spare	181 GB 1× · 107 GB free
Laguna-S 2.1 Poolside	118B 8B active	282 GB 1× · 5.8 GB free	155 GB 1× · 133 GB free	90 GB 1× · 198 GB free
gpt-oss-120b OpenAI	117B 5.1B active	— Not released	— Not released	78 GB 1× · 210 GB free
Qwen3.6 35B-A3B Alibaba	35B 3B active	86 GB 1× · 202 GB free	47 GB 1× · 241 GB free	28 GB 1× · 260 GB free
Qwen3 32B Alibaba	32.8B	79 GB 1× · 209 GB free	43 GB 1× · 245 GB free	25 GB 1× · 263 GB free
Gemma 4 31B Google	31.3B	75 GB 1× · 213 GB free	41 GB 1× · 247 GB free	24 GB 1× · 264 GB free
GLM-4.7 Flash Z.ai	31.2B	75 GB 1× · 213 GB free	41 GB 1× · 247 GB free	24 GB 1× · 264 GB free
Qwen3.6 27B Alibaba	27.8B	67 GB 1× · 221 GB free	37 GB 1× · 251 GB free	21 GB 1× · 267 GB free
Mistral Small 3.2 24B Mistral AI	24B	58 GB 1× · 230 GB free	32 GB 1× · 256 GB free	18 GB 1× · 270 GB free
gpt-oss-20b OpenAI	21B 3.6B active	— Not released	— Not released	17 GB 1× · 271 GB free

Estimates: model weights plus 20% for KV cache and runtime overhead, at short context lengths. Long contexts and large batches need more. Rows marked “tight” hold the weights but not that full margin. A ×N figure is the total VRAM across N of these GPUs and makes no claim about interconnect throughput. How we estimate this · Start from a model instead

Want more from this page?

See incorrect data?

Frequently asked questions

How much VRAM does the Instinct MI350X have?

The Instinct MI350X comes with 288 GB of VRAM, which determines the largest models and batch sizes it can hold in memory for training and inference.

Which LLMs can I run on the Instinct MI350X?

With 288 GB of VRAM, a single Instinct MI350X can serve models such as DeepSeek V4 Flash, Solar Open2 250B, Qwen3 235B-A22B, Laguna-S 2.1. The table above lists the estimated VRAM for each model at 16-bit, 8-bit and 4-bit precision, including the multi-GPU configurations needed for larger models.