NV NVIDIA Consumer Discrete GPU

NVIDIA GeForce RTX 4070 Ti Super

Runs downloadable LLMs up to ~19.5B at Q4, ~109.5-130.5 tok/s on an 8B model.

NVIDIA GeForce RTX 4070 Ti Super
16GB
VRAM
672
GB/s bandwidth
19.5B
Max model (Q4)
$799
MSRP

What you can run

At 8k context, FP16 KV. Full-speed tok/s holds up to "max ctx"; beyond it the context spills to system RAM (CPU/PCIe offload) and drops to the degraded figure.

Model sizeFP16Q8Q4
3B 81.8-97.4 t/s
≤110k ctx · then ~4.9-5.8 (slow)
163.5-194.9 t/s
≤156k ctx · then ~9.7-11.6 (slow)
292-348 t/s
≤176k ctx · then ~17.4-20.7 (slow)
7-8B won't run 61.3-73.1 t/s
≤40k ctx · then ~3.7-4.4 (slow)
109.5-130.5 t/s
≤67k ctx · then ~6.5-7.8 (slow)
13B won't run 37.7-45 t/s
≤1k ctx · then ~2.2-2.7 (slow)
67.4-80.3 t/s
≤36k ctx · then ~4-4.8 (slow)
32B won't run won't run won't run
70B won't run won't run won't run
120B won't run won't run won't run
400B won't run won't run won't run

Estimate: tok/s ≈ bandwidth ÷ (params × bytes/param) × efficiency (NVIDIA). Cyan = fits ~14.7 GB usable. Beyond "ctx", KV spills to system RAM (CPU/PCIe offload) (rough degraded speed shown).

What else you need for a full build

Minimum supporting components for a desktop built around the NVIDIA GeForce RTX 4070 Ti Super. Guidance — exact needs depend on your case, other parts and how much you offload to system RAM.

Power supply
≥ 750W · 80+ Gold
Headroom over the 285W card plus the rest of the build.
System RAM
≥ 16 GB · 32 GB recommended
At least the 16 GB of VRAM; 2× lets you CPU-offload models larger than VRAM.
Motherboard
1× free PCIe x16 slot (4.0/5.0)
A full-length x16 electrical slot; PCIe 4.0 is ample, 5.0 future-proofs.
CPU
Any modern multi-core
The GPU runs inference; more cores and memory channels only matter when offloading to RAM.
Storage
≥ 1 TB NVMe SSD
Weights are large — a 70B model at Q4 is ~40 GB; budget for a few.

Run popular models on the NVIDIA GeForce RTX 4070 Ti Super

Videos about the NVIDIA GeForce RTX 4070 Ti Super

Should You Buy nVidia RTX 4070ti Super GPU for Local AI? Kimi k2.6 agents?

Mistral 7B LLM AI Leaderboard: Unboxing an Nvidia RTX 4070Ti Super 16GB and giving it run!

More hardware like the NVIDIA GeForce RTX 4070 Ti Super

About the NVIDIA GeForce RTX 4070 Ti Super

NVIDIA GeForce RTX 4070 Ti Super is a discrete GPU from NVIDIA on Ada Lovelace, 2024. For local genAI the numbers that matter are its 16 GB (how big a model + context fits) and 672 GB/s bandwidth (generation speed). Plan for a ~750W power supply.

Frequently asked questions

What size LLM can the NVIDIA GeForce RTX 4070 Ti Super run?

With ~14.7 GB usable it runs up to ~19.5B at Q4 (~6B at FP16) at 8k context. Bigger spills into system RAM (CPU/PCIe offload) and slows sharply.

How many tokens/sec does the NVIDIA GeForce RTX 4070 Ti Super generate?

Generation is bandwidth-bound. At 672 GB/s an 8B model at Q4 runs ~109.5-130.5 tok/s (estimate).

Why does long context slow it down?

The KV cache for the context must also fit fast memory; once weights+KV exceed 14.7 GB it spills to system RAM (CPU/PCIe offload) and throughput collapses. Each row shows the max usable context.

What else do I need to run the NVIDIA GeForce RTX 4070 Ti Super?

Besides the card: a ≥750W 80+ Gold power supply, at least 16GB of system RAM (2× the 16GB VRAM is ideal for CPU offload), a free PCIe x16 slot, any modern multi-core CPU, and a ≥1TB NVMe SSD for model weights.