NVIDIA RTX 4090

24 GB VRAM450 W TDP10 providers79 offerings19 regions

The NVIDIA RTX 4090 is a consumer-grade GPU built on the Ada Lovelace architecture, featuring 24 GB of memory with 1,008 GB/s of memory bandwidth and a 450 W TDP. It delivers 82.6 TFLOPS of FP32 performance. Currently available from 10 providers starting at $0.13/GPU/hour, with a market median of $0.56/GPU/hour across 79 configurations.

RTX 4090 · Founders Edition reference

Representative design. Provider hardware may differ.

Cheapest

$0.130/hr

Vast.ai

Median

$0.560/hr

across 79 offerings

Most expensive

$1.270/hr

Vast.ai

90-day trend

+14.3%

median rate

Hardware specifications

Same across all providers

Memory

24 GB

GDDR6X

What LLMs fit? ↓

TDP

450 W

Memory bandwidth

1,008 GB/s

Architecture

Ada Lovelace

FP32 performance

82.6 TFLOPS

Brand

NVIDIA

Series

GeForce RTX

Market price history

Median price across all providers

Prices updated 10 minutes ago

11 offerings from 10 providers

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

Provider	Count	vCPU	RAM	Region	Per GPU hour	Total/hr	Action
RunpodPromoted Up to$500creditAffiliate promotion — we may earn a commission.	×1	—	—	--	$0.890	$0.89	Launch
Vast.aiReferral linkTrendingCheapest	×1	0	16 GB	+14	From$0.130 -20.9% 30d	From$0.13	Launch
Theta EdgeCloudTrending	×1	20	47 GB		$0.250 +25.0% 30d	$0.25	Launch
Akash Network Up to$100credit	×1	—	—		$0.280	$0.28	Launch
LiumReferral link Up to$5credit	×8	256	472 GB		From$0.400 +47.1% 30d	From$3.20	Launch
Omega Gradient	×8	88	200 GB		From$0.420 +2.9% 30d	From$3.36	Launch
Sesterce CloudReferral link	×8	88	200 GB		From$0.440 +20.0% 30d	From$3.52	Launch
Runcrate	×1	12	70 GB		$0.520 +18.2% 30d	From$0.52	Launch
Novita AIReferral link	×1	16	125 GB		$0.670	$0.67
RunpodReferral linkTrending Up to$500credit	×1	—	—	--	$0.890 0.0% 30d	$0.89	Launch
EmpirioLabs AI	×1	80	63 GB		From$0.900 +50.0% 30d	From$0.90	Launch
Daytona Up to$200credit	×1	Billed separately	Billed separately	--	$0.990 0.0% 30d	From$0.99	Launch

LLMs that fit on the NVIDIA RTX 4090

Estimated VRAM for popular open-weight LLMs at 16-bit (FP16/BF16), 8-bit (FP8/INT8) and 4-bit (INT4/MXFP4) precision, against 24 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 Barely fits on 8×
Solar Open2 250B Upstage	250B 15B active	— Not released	330 GB 14× · 5.6 GB free or 12× · 4% spare	192 GB 11× · 72 GB free or 8× · 19% spare
Qwen3 235B-A22B Alibaba	235B 22B active	— Not released	310 GB 14× · 26 GB free or 11× · 2% spare	181 GB 8× · 11 GB free
Laguna-S 2.1 Poolside	118B 8B active	282 GB 12× · 5.8 GB free or 11× · 12% spare	155 GB 8× · 37 GB free or 6× · 11% spare	90 GB 4× · 5.7 GB free
gpt-oss-120b OpenAI	117B 5.1B active	— Not released	— Not released	78 GB 4× · 18 GB free or 3× · 10% spare
Qwen3.6 35B-A3B Alibaba	35B 3B active	86 GB 4× · 9.7 GB free or 3× · 0% spare	47 GB Barely fits on 2×	28 GB 2× · 20 GB free or 1× · 4% spare
Qwen3 32B Alibaba	32.8B	79 GB 4× · 17 GB free or 3× · 9% spare	43 GB 2× · 4.8 GB free	25 GB 2× · 23 GB free or 1× · 14% spare
Gemma 4 31B Google	31.3B	75 GB 4× · 21 GB free or 3× · 15% spare	41 GB 2× · 6.7 GB free	24 GB 2× · 24 GB free or 1× · 19% spare
GLM-4.7 Flash Z.ai	31.2B	75 GB 4× · 21 GB free or 3× · 15% spare	41 GB 2× · 6.8 GB free	24 GB Barely fits on 1×
Qwen3.6 27B Alibaba	27.8B	67 GB 3× · 5.3 GB free	37 GB 2× · 11 GB free	21 GB 1× · 2.7 GB free
Mistral Small 3.2 24B Mistral AI	24B	58 GB 3× · 14 GB free	32 GB 2× · 16 GB free	18 GB 1× · 5.6 GB free
gpt-oss-20b OpenAI	21B 3.6B active	— Not released	— Not released	17 GB 1× · 7.5 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

See more GPUs like this in

Best bang for buck

Want more from this page?

See incorrect data?

Frequently asked questions

How much does the RTX 4090 cost per hour?

The RTX 4090 starts at $0.13 per GPU-hour, with a median of $0.56 across 79 available configurations. Actual pricing depends on the provider, region, and whether you rent on-demand, spot or reserved.

Which cloud providers offer the RTX 4090?

11 providers currently list the RTX 4090: Runpod, Daytona, Sesterce Cloud, Theta EdgeCloud, Akash Network, Omega Gradient and 5 more. Compare their hourly pricing, regions and machine specs in the table above.

How much VRAM does the RTX 4090 have?

The RTX 4090 comes with 24 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 RTX 4090?

With 24 GB of VRAM, a single RTX 4090 can serve models such as GLM-4.7 Flash, Qwen3.6 27B, Mistral Small 3.2 24B, gpt-oss-20b. 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.

What's the biggest LLM I can run on multiple RTX 4090 GPUs?

The largest models we track that fit across several RTX 4090 cards are DeepSeek V4 Flash on 8 GPUs at 4-bit, Solar Open2 250B on 11 GPUs at 4-bit, Qwen3 235B-A22B on 8 GPUs at 4-bit. A multi-GPU configuration means you rent and pay for every card for the whole run, and splitting a model across cards costs throughput as well: the GPUs exchange activations on every token, so performance depends on how fast they are wired together. We do not track interconnect, so check the provider's node specification. The table above gives the smallest configuration for each model at 16-bit, 8-bit and 4-bit precision.

Where is the RTX 4090 cheapest?

The lowest on-demand price we currently track for the RTX 4090 is $0.13 per GPU-hour at Vast.ai. Because providers update pricing and availability frequently, check the live table above before you launch.