GeForce RTX 4070 Ti SUPER Price, 16GB VRAM & Local LLMs
Price alert
We'll email you when the eBay (US) price reaches your target.
NVIDIA's RTX 4070 Ti SUPER (2024) is an upper-midrange gaming card and the only 16GB model in the 4070 family, so it fits 12B-class models even at 8-bit. Sellers often shorten it to “4070 TiS” or “4070 TS” in listings; don't confuse it with the 12GB 4070 Ti.
eBay
GeForce RTX 4070 Ti SUPER price history
eBay
As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.
GeForce RTX 4070 Ti SUPER VRAM & full specs
- Memory16 GB · GDDR6X
- Bandwidth672 GB/s
- FP16 / BF16176.4 TFLOPS
- FP8 / INT8352.8 / 352.8 TOPS
- Bus width
- 256-bit
- Architecture
- Ada Lovelace (AD103)
- Cores
- 8,448
- Tensor cores
- 264
- Ecosystem
- CUDA
- Interface
- PCIe 4.0 x16
- NVLink
- Not supported
- Power
- 285 W
- Power connector
- 16-pin 12VHPWR（盒内附 2× 8-pin 转接线）
- Type
- Consumer
- Launch date
- Jan 24, 2024
- MSRP
- $799
Best GeForce RTX 4070 Ti SUPER deals
Powered by VRAMGlass real-time price tracking — handpicking the best-value hardware deals across the web
As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.
UsedMSI Nvidia RTX 4070 Ti Super 16GB Graphics Card GPU PCTop Rated$1,050Buy on eBay
NewZOTAC Gaming GeForce RTX 4070 Ti Super AMP Holo 16 GB GDDR6X Graphics Ca$1,099+ shippingBuy on eBay
UsedZotac GeForce RTX 4070 Ti Super Trinity RGB Black 16GB GDDR6X Graphics Card$1,090+$10 shippingBuy on eBay
UsedDell Nvidia GeForce RTX 4070 Ti Super 16GB GDDR6X Gaming Graphics CardTop Rated$1,160Buy on eBay
UsedNvidia ACER Predator Orion RTX 4070 Ti Super 16GB OEM Graphics CardTop Rated$1,160Buy on eBay
UsedGIGABYTE Windforce OC GeForce RTX 4070 Ti Super 16GB - Graphics Card - ExcellentTop Rated$1,200Buy on eBay
As an Amazon Associate I earn from qualifying purchases. The price you pay stays the same.
UsedGaming RTX 4070 TI Super 16G Ventus 3X OC Graphics Card (NVIDIA RTX 4070 Ti Super, 256-Bit, Boost Clock: 2655 MHz, 16GB GDRR6X 21 Gbps, HDMI/DP, Ada Lovelace Architecture) (Renewed)$1,350Buy on Amazon
Newmsi Gaming RTX 4070 Ti Super 16G Ventus 2X OC Graphics Card (NVIDIA RTX 4070 Ti Super, 256-Bit, Extreme Clock: 2655 MHz, 16GB GDRR6X 21 Gbps, HDMI/DP, Ada Lovelace Architecture)$1,499+$6 shippingBuy on Amazon
UsedGeForce RTX 4070 Ti Super Eagle OC 16G Graphics Card, 3X WINDFORCE Fans, 16GB 256-bit GDDR6X, GV-N407TSEAGLE OC-16GD Video Card (Renewed), Black$1,600Buy on Amazon
What LLMs can a single GeForce RTX 4070 Ti SUPER run?
A single RTX 4070 Ti SUPER can run models up to 12B at 8-bit, with Gemma 4 12B at about 38 tokens/s. Dense models of 27B and up don't fit at 4-bit and have to drop to 2-bit; for MoE, Qwen3.6 35B A3B at 2-bit fits entirely on the card and is the fastest option it has.
| Model | Size | Runs? | Speed t/s | Max context |
|---|---|---|---|---|
 Gemma 4 E4B8B | 3.8 GB | Runs | 112 | 128K |
 Qwen3.5 9B9.7B | 4.1 GB | Runs | 104 | 256K |
| Gemma 4 12B12B | 4.7 GB | Runs | 91.2 | 256K |
| Gemma 4 26B A4B25.2B (3.8B active) | 10.5 GB | Runs | 123 | 256K |
| Qwen3.6 27B27.8B | 11.8 GB | Runs | 40.3 | 64K |
| Qwen3.8 27B27.8B | 9.8 GB | Runs | 47.8 | 93K |
 GLM 4.7 Flash30B (3B active) | 11.9 GB | Runs | 134.6 | 75K |
| Gemma 4 31B30.7B | 11.8 GB | Runs | 38.8 | 41K |
| Qwen3.6 35B A3B36B (3B active) | 12.3 GB | Runs | 145.9 | 181K |
| Model | Size | Runs? | Speed t/s | Max context |
|---|---|---|---|---|
| Gemma 4 E4B8B | 5 GB | Runs | 88.9 | 128K |
| Qwen3.5 9B9.7B | 5.7 GB | Runs | 78.6 | 256K |
| Gemma 4 12B12B | 7.1 GB | Runs | 64 | 256K |
| Gemma 4 26B A4B25.2B (3.8B active) | 16.9 GB | Needs RAM offload | 37.4 | 256K |
| Qwen3.6 27B27.8B | 16.8 GB | Too big | — | — |
| Qwen3.8 27B27.8B | 16.5 GB | Too big | — | — |
| GLM 4.7 Flash30B (3B active) | 18.3 GB | Needs RAM offload | 35.9 | 198K |
| Gemma 4 31B30.7B | 18.3 GB | Too big | — | — |
| Qwen3.6 35B A3B36B (3B active) | 22.1 GB | Needs RAM offload | 50.3 | 256K |
| Model | Size | Runs? | Speed t/s | Max context |
|---|---|---|---|---|
| Gemma 4 E4B8B | 8.2 GB | Runs | 57.3 | 128K |
| Qwen3.5 9B9.7B | 9.5 GB | Runs | 49.8 | 196K |
| Gemma 4 12B12B | 12.7 GB | Runs | 37.7 | 183K |
| Gemma 4 26B A4B25.2B (3.8B active) | 26.9 GB | Needs RAM offload | 25.3 | 256K |
| Qwen3.6 27B27.8B | 28.6 GB | Too big | — | — |
| Qwen3.8 27B27.8B | 29 GB | Too big | — | — |
| GLM 4.7 Flash30B (3B active) | 31.8 GB | Needs RAM offload | 22.4 | 198K |
| Gemma 4 31B30.7B | 32.6 GB | Too big | — | — |
| Qwen3.6 35B A3B36B (3B active) | 36.9 GB | Needs RAM offload | 33.4 | 256K |
RunsFits fully in VRAM; speed is estimated from VRAM bandwidth.
Needs RAM offloadMoE only: expert weights sit in system RAM, so speed is estimated from 70 GB/s RAM bandwidth.
Too bigWon’t fit on one card at this quantization, even with RAM offload.
Full calculation rules are on the Methodology page.
Who it suits
Good fit
- You want models up to 12B at 8-bit
Gemma 4 12B at 8-bit (12.7 GB) runs at about 38 tokens/s with up to 183K of context, versus about 64 at 4-bit
- Running 30B-class MoE models at 2-bit entirely on the card
Qwen3.6 35B A3B runs at about 146 tokens/s with 181K of context, nearly three times as fast as at 4-bit with its experts in system RAM
- You want generation speed close to a 4080 with lower power draw
It has the same 16GB, the 4080's bandwidth is only 7% higher, and this card draws 285W, 35W less than the 4080
Not a good fit
- You want 27B-class dense models at 4-bit
The weights alone start at 16.5 GB, so once you add context they won't fit in 16GB of VRAM; they do run at 2-bit, but Gemma 4 31B is left with only 41K of context
- Running 100B-plus models across several cards
It takes 8 cards to fit Qwen3.8 Flash Next at 2-bit, and finding a motherboard and power supply for that many is hard
Before you buy
Make sure it's the Ti SUPER with 16GB
The 4070 Ti and 4070 Ti SUPER differ by a single word: the first has 12GB on a 192-bit bus, the second 16GB on a 256-bit bus. Sellers often shorten the SUPER to “4070 TiS” or “4070 TS” in listings, and one listed only as “4070 Ti” is most likely the 12GB card. Before ordering, ask the seller for a GPU-Z screenshot and check the VRAM and bus width.
No Founders Edition, and seat the power plug fully
NVIDIA made no Founders Edition of the 4070 Ti SUPER, so look up size and cooling for the specific model; NVIDIA recommends a 700W power supply for the system. Most cards have a single 16-pin power connector, with an adapter for two 8-pin cables in the box. As with the 4080 and 4090, push the plug in all the way before you power on.
Same chip as the 4080
It uses a cut-down version of the 4080's AD103 chip with the same 256-bit memory bus, so its generation speed is very close to the 4080's. The difference is compute: the 4080 has about 10% more, which makes it a bit faster when reading in a very long document at once.
Compared with similar GPUs
GeForce RTX 4070 Ti SUPER vs GeForce RTX 3090, GeForce RTX 5070 Ti, and GeForce RTX 4080: price, VRAM, bandwidth, compute and local LLM speed compared.
| Spec | GeForce RTX 4070 Ti SUPER | GeForce RTX 3090 | GeForce RTX 5070 Ti | GeForce RTX 4080 |
|---|---|---|---|---|
| Memory | 16 GB | 24 GB50% more | 16 GBSame | 16 GBSame |
| Bandwidth | 672 GB/s | 936 GB/s39% higher | 896 GB/s33% higher | 717 GB/s7% higher |
| BF16 / FP16 compute | 176 TFLOPS | 142 TFLOPS19% lower | 176 TFLOPSSame | 195 TFLOPS11% higher |
| INT8 compute | 353 TOPS | 285 TOPS19% lower | 352 TOPSSame | 390 TOPS11% higher |
| Qwen3.8 27B speed (2-bit) | 47.8 t/s | 64.8 t/s36% faster | 62.3 t/s30% faster | 50.7 t/s6% faster |
| Largest dense model (4-bit) | 23B | 37B61% more | 23BSame | 23BSame |
| Power | 285 W | 350 W23% higher | 300 W5% higher | 320 W12% higher |
| GPU interconnect | Not supported | NVLink 112.5 GB/s | Not supported | Not supported |
| Interface | PCIe 4.0 x16 | PCIe 4.0 x16 | PCIe 5.0 x16 | PCIe 4.0 x16 |
| Released | 2024-01 | 2020-09 | 2025-02 | 2022-11 |
| eBay median |
As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.
Amazon.jp
Xianyu