NVIDIA GeForce RTX 3070 vs NVIDIA GeForce RTX 4070 Ti SUPER — Specs & Performance Compared

Verdict: The NVIDIA GeForce RTX 4070 Ti SUPER is ahead of the NVIDIA GeForce RTX 3070 on all 7 shared benchmarks on file, by 23% to 90%, including 64% in 3DMark Time Spy (22,990 vs 13,997). On llama3.1:8b, the NVIDIA GeForce RTX 3070 generates 73.7 tokens/s (per Patrick Hughes (bmdpat.com)) against 53.9 for the NVIDIA GeForce RTX 4070 Ti SUPER (per LocalScore (Mozilla Builders)), best community run on file for each.

Side-by-side GPU comparison of the NVIDIA GeForce RTX 3070 and the NVIDIA GeForce RTX 4070 Ti SUPER. Released 2020 (NVIDIA GeForce RTX 3070), 2024 (NVIDIA GeForce RTX 4070 Ti SUPER). MSRP range: $499 · $799. Specs, synthetic benchmarks, AI inference tok/s, and Amazon availability below; each column header links to that part's benchmark page.

NVIDIA · Ampere

NVIDIA GeForce RTX 3070

8 GB VRAM220W TDP$499 MSRP2020

Full NVIDIA GeForce RTX 3070 benchmarks →

NVIDIA · Ada Lovelace

NVIDIA GeForce RTX 4070 Ti SUPER

16 GB VRAM285W TDP$799 MSRP2024

Full NVIDIA GeForce RTX 4070 Ti SUPER benchmarks →

*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

Spec	NVIDIA GeForce RTX 3070	NVIDIA GeForce RTX 4070 Ti SUPER
vram gb	8	16
vram type	GDDR6	GDDR6X
tdp w	220	285
cuda cores	5888	8448

Synthetic benchmarks

Each row says whether a higher or a lower number is better (ranks and times are lower-is-better). Scores come from WorthPlaying, KitGuru, DigiStatement, Guru3D, 3DMark, Overclocking.com, VideoCardz, WCCFTech, TopCPU.net, UL Benchmarks and TechPowerUp; each row links its sources, and where they differ the cell shows their median. Empty cells mean we don't yet have that benchmark on file for the SKU.

Benchmark	NVIDIA GeForce RTX 3070	NVIDIA GeForce RTX 4070 Ti SUPER
3DMark Fire Strike (higher is better) Source: WorthPlaying	—	42,724 points
3DMark Fire Strike Extreme (higher is better) Source: KitGuru, WorthPlaying, DigiStatement	17,115 points	26,435 points
3DMark Fire Strike Ultra (higher is better) Source: Guru3D, WorthPlaying	8,749 points	14,209 points
3DMark Port Royal (higher is better) Source: KitGuru, 3DMark, Overclocking.com, DigiStatement, VideoCardz, WCCFTech	8,324 points	15,798 points
3DMark Speed Way (higher is better) Source: 3DMark, Overclocking.com, TopCPU.net, WCCFTech	3,421 points	6,395 points
3DMark Steel Nomad (higher is better) Source: UL Benchmarks	3,181 points	—
3DMark Steel Nomad (DX12) graphics score (higher is better) Source: 3DMark	3,182 points	—
3DMark Steel Nomad Graphics (higher is better) Source: UL Benchmarks	3,180 points	—
3DMark Time Spy (higher is better) Source: KitGuru, 3DMark, WorthPlaying, DigiStatement, Overclocking.com, VideoCardz	13,997 points	22,990 points
3DMark Time Spy (Graphics) (higher is better) Source: WCCFTech	—	23,983 points
3DMark Time Spy Extreme (higher is better) Source: TechPowerUp, KitGuru, WorthPlaying, DigiStatement, Overclocking.com	6,907 points	11,170 pts
3DMark Time Spy Extreme (Graphics) (higher is better) Source: WCCFTech	—	11,981 points

AI inference (tokens/sec)

Top community-reported tokens-per-second results for popular LLMs whose weights fit the card's memory, each linked to its source. See the per-SKU benchmark page for the full set including quantization and prompt length.

NVIDIA GeForce RTX 3070

llama2:7b: 78.7 tok/s (llama.cpp GitHub Discussions)
llama2:7b: 78.7 tok/s (llama.cpp GitHub Discussion #10879)
llama3.1:8b: 73.7 tok/s (Patrick Hughes (bmdpat.com))

All NVIDIA GeForce RTX 3070 AI benchmarks →

NVIDIA GeForce RTX 4070 Ti SUPER

llama3.2:1b: 161.0 tok/s (LocalScore (Mozilla Builders))
llama2:7b: 132.3 tok/s (llama.cpp GitHub (CUDA Scoreboard))
llama2:7b: 132.3 tok/s (llama.cpp GitHub (CUDA performance scoreboard))

All NVIDIA GeForce RTX 4070 Ti SUPER AI benchmarks →

Quick answers

Which is faster, the NVIDIA GeForce RTX 3070 or the NVIDIA GeForce RTX 4070 Ti SUPER?

The NVIDIA GeForce RTX 4070 Ti SUPER is ahead of the NVIDIA GeForce RTX 3070 on all 7 shared benchmarks on file, by 23% to 90%, including 64% in 3DMark Time Spy (22,990 vs 13,997).

Which is better for running local LLMs?

On llama3.1:8b, the NVIDIA GeForce RTX 3070 generates 73.7 tokens/s (per Patrick Hughes (bmdpat.com)) against 53.9 for the NVIDIA GeForce RTX 4070 Ti SUPER (per LocalScore (Mozilla Builders)), best community run on file for each.

What did the NVIDIA GeForce RTX 3070 and the NVIDIA GeForce RTX 4070 Ti SUPER cost at launch?

The NVIDIA GeForce RTX 3070 launched at a $499 MSRP and the NVIDIA GeForce RTX 4070 Ti SUPER at $799. Street prices drift from launch pricing; the Where to buy section links current listings.