RTX 3090 vs RTX 4090
Side-by-side comparison of the NVIDIA RTX 3090 and the NVIDIA RTX 4090 for AI inference workloads.
Specifications
Pricing
RTX 3090
| Provider | On-Demand | Reserved | Spot |
|---|---|---|---|
| runpod | $0.54/hr | - | $0.34/hr |
| vast ai | $0.36/hr | - | $0.20/hr |
| tensordock | $0.29/hr | - | $0.19/hr |
RTX 4090
| Provider | On-Demand | Reserved | Spot |
|---|---|---|---|
| runpod | $1.10/hr | - | $0.79/hr |
| lambda | $0.89/hr | - | - |
| vast ai | $0.74/hr | - | $0.44/hr |
| tensordock | $0.69/hr | - | $0.44/hr |
| fluidstack | $0.59/hr | - | $0.39/hr |
Cheapest available rate: RTX 3090 at $0.19/hr vs RTX 4090 at $0.39/hr — RTX 3090 is +105% cheaper
Efficiency Metrics
TFLOPS / Watt
0.1
RTX 3090
0.4
RTX 4090
BF16
VRAM / Dollar
126.3
RTX 3090
61.5
RTX 4090
GB/$/hr
Bandwidth / Watt
2.7
RTX 3090
2.2
RTX 4090
GB/s/W
Models (FP16, 1 GPU)
124.0
RTX 3090
124.0
RTX 4090
Model Compatibility (FP16, Single GPU)
Only on RTX 3090 (0)
None
Both (124)
- Yi 1.5 9B
- Yi Coder 9B
- GTE Qwen2 7B
- Marco O1
- Qwen 2 Audio 7B
- Qwen 2.5 3B
- OLMo 2 7B
- OpenELM 3B
- BGE Large EN v1.5
- BGE M3
- StarCoder2 3B
- StarCoder2 7B
- Aya 23 8B
- Command R 7B
- DeepSeek Coder 6.7B
- DeepSeek Math 7B
- DeepSeek R1 Distill 8B
- Falcon 7B
- Gemma 1.1 2B
- Gemma 2 2B
- +104 more
Only on RTX 4090 (0)
None
Summary
The RTX 3090 (ampere generation) offers 24GB of GDDR6X with 36 BF16 TFLOPS and 936 GB/s memory bandwidth at 350W TDP.
The RTX 4090 (ada generation) offers 24GB of GDDR6X with 165 BF16 TFLOPS and 1,008 GB/s memory bandwidth at 450W TDP.
From a cost perspective, the RTX 3090 is more affordable at $0.19/hr vs $0.39/hr for the RTX 4090.