Skip to main content

NVIDIA GeForce RTX 4090 vs NVIDIA GeForce RTX 5090 — Specs & Performance Compared

Side-by-side comparison of 2 GPU parts: NVIDIA GeForce RTX 4090 vs NVIDIA GeForce RTX 5090. Released 2022 (NVIDIA GeForce RTX 4090), 2025 (NVIDIA GeForce RTX 5090). MSRP range: $1,599 · $1,999. Specs, synthetic benchmarks, AI inference tok/s, and Amazon availability — click into any column header for the per-SKU benchmark page.

Verdict: The NVIDIA GeForce RTX 5090 is ahead of the NVIDIA GeForce RTX 4090 on all 4 shared benchmarks on file, by 2% to 60% (60% in 3DMark Port Royal). On llama3.1:70b, the NVIDIA GeForce RTX 4090 generates 19.1 tokens/s against 18.5 for the NVIDIA GeForce RTX 5090 (best community run on file for each).

vs

*Price sourced from Amazon.com. Price and availability subject to change.

Where to buy

A listing for each part; the current price and stock are on the retailer's page.

Specs side-by-side

SpecNVIDIA GeForce RTX 4090NVIDIA GeForce RTX 5090
vram gb 2432
vram type GDDR6XGDDR7
tdp w 450575
cuda cores 1638421760

Synthetic benchmarks

Each row says whether a higher or a lower number is better (ranks and times are lower-is-better). Per-row scores come from public benchmark databases (PassMark, TechPowerUp, Geekbench, Cinebench). Empty cells mean we don't yet have that benchmark on file for the SKU.

BenchmarkNVIDIA GeForce RTX 4090NVIDIA GeForce RTX 5090
3DMark Port Royal (higher is better) 25,820 points41,238 points
3DMark Speed Way (higher is better) 10,527 points15,488 points
3DMark Steel Nomad (higher is better) 27,805 points11,906 points
3DMark Steel Nomad (4K) (higher is better) —14,133 points
3DMark Steel Nomad Light (higher is better) 11,287 points—
3DMark Time Spy (higher is better) 37,979 points39,365 pts
3DMark Time Spy Extreme (higher is better) 29,321 pts—
3DMark Time Spy Extreme (Graphics) (higher is better) 19,000 pts—
3DMark Time Spy GPU (higher is better) 35,200 pts—
Deep Learning Benchmark (ResNet-50) (higher is better) 128,000 images/sec—
PassMark G2D Mark (higher is better) 1,302 pts1,412 pts
PassMark G3D Mark (higher is better) 38,066 pts38,935 pts

AI inference (tokens/sec)

Top community-reported tokens-per-second results for popular LLMs. See the per-SKU benchmark page for the full set including quantization, prompt length, and source citations.

NVIDIA GeForce RTX 4090

  • llama3.1:8b: 2550.0 tok/s
  • qwen3:32b: 650.0 tok/s
  • llama3:8b: 440.0 tok/s

All NVIDIA GeForce RTX 4090 AI benchmarks →

NVIDIA GeForce RTX 5090

  • Qwen2.5-Coder-7B-Instruct: 5841.0 tok/s
  • Llama 3.1 8B FP8: 420.0 tok/s
  • Llama 2 7B: 263.6 tok/s

All NVIDIA GeForce RTX 5090 AI benchmarks →

Quick answers

Which is faster, the NVIDIA GeForce RTX 4090 or the NVIDIA GeForce RTX 5090?

The NVIDIA GeForce RTX 5090 is ahead of the NVIDIA GeForce RTX 4090 on all 4 shared benchmarks on file, by 2% to 60% (60% in 3DMark Port Royal).

Which is better for running local LLMs?

On llama3.1:70b, the NVIDIA GeForce RTX 4090 generates 19.1 tokens/s against 18.5 for the NVIDIA GeForce RTX 5090 (best community run on file for each).

What did the NVIDIA GeForce RTX 4090 and the NVIDIA GeForce RTX 5090 cost at launch?

The NVIDIA GeForce RTX 4090 launched at a $1,599 MSRP and the NVIDIA GeForce RTX 5090 at $1,999. Street prices drift from launch pricing; the Where to buy section links current listings.

Keep exploring

More guides & deep dives from the SpecPicks archive

Browse all articles & guides →

More reviews from the SpecPicks archive

Browse all reviews →

More buying guides from SpecPicks

Browse all buying guides →

Hardware benchmark data on SpecPicks

All benchmarks →