← Back to GPUs

NVIDIA · RTX 30
NVIDIA GeForce RTX 3090 Ti
$1000$1999 MSRP
The NVIDIA GeForce RTX 3090 Ti pushed the Ampere architecture to its limits with 24GB of GDDR6X at higher bandwidth than the standard 3090. On the used market, it offers slightly faster AI inference than the 3090 at a modest price premium. The 450W TDP is aggressive, requiring a robust PSU and good airflow. A solid used-market AI pick for those who want the fastest 24GB Ampere option.
Best ForUsed market AI builds wanting the fastest 24GB Ampere card
VerdictFaster than the 3090 but much hungrier — worth the premium if power isn't a concern.
AI
8/10
Gaming
7/10
Specifications
VRAM24GB GDDR6X
Memory Bandwidth1008 GB/s
CUDA Cores10,752
Boost Clock1860 MHz
TDP450W
Power Connector1x 16-pin
Length336mm
Form FactorTriple Slot
Release Year2022
AI Capabilities
Sweet Spot24GB VRAM
The professional standard. Handles most models with smart quantization.
Can run (Q4 quantized)
Llama 3.1 8BQwen 2.5 32BQwen 2.5 14BMistral 7BFLUX.1 DevStable Diffusion XLStable Diffusion 3.5 LargeHunyuanVideoCogVideoX-5BMochi 1LTX VideoStable Video DiffusionWan Video 14BCodestral 22BQwen 2.5 Coder 32BLLaVA 1.6 34BAlphaFold 2ESMFold (ESM-2 15B)ESM-2 3BscGPTRFdiffusionFine-tune Llama 8BTrain SDXL LoRATrain FLUX LoRA
Recommended system RAM for AI: 48GB+ (2x GPU VRAM for model overflow)
Performance Estimates
Estimated tokens/sec for LLM inference based on 1008 GB/s memory bandwidth — not hardware benchmarks. Methodology · What is Q4/Q8?
Llama 3.1 8B8B
FP16~31-38 tok/sFastQwen 2.5 32B32B
Q4~27-34 tok/sUsableQwen 2.5 14B14B
Q8~37-46 tok/sFastMistral 7B7B
FP16~35-44 tok/sFastCodestral 22B22B
Q8~24-29 tok/sUsableQwen 2.5 Coder 32B32B
Q4~27-34 tok/sUsablePros
- +24GB with fast bandwidth
- +Good used prices for AI
- +Faster than 3090
Cons
- -Extremely power hungry
- -16-pin adapter
- -Runs very hot
aigaming
Where to Buy
Will It Run?
Llama 3.1 8B8B
FP16Qwen 2.5 32B32B
Q4Qwen 2.5 14B14B
Q8Mistral 7B7B
FP16FLUX.1 Dev12B
Q8Stable Diffusion XL6.6B
FP16Stable Diffusion 3.5 Large8B
FP16HunyuanVideo13B
Q8