112 FPS average at 1440p across 15 games, first-party measured · smooth 100+ FPS 1440p gaming across demanding AAA titles
112 Avg FPS ✓ Measured
These are our own numbers. 15 games, one bench, one card, with the capture published for each result. Across our 15-game suite at 1440p, GeForce RTX 5070 averages 112 FPS. The ceiling is Counter-Strike 2 at 234 FPS; the floor is Clair Obscur: Expedition 33 at 49 FPS, a 4.8x spread that tells you how engine-dependent this class of card is. In practical terms that means smooth 100+ FPS 1440p gaming across demanding AAA titles. Under load we logged a peak of 251W against the 250W rating.
Black Desert Online was the pacing outlier in my runs (1% low at 47% of average). Tarkov used 11.51GB of the 12GB. This card is closer to its VRAM ceiling than NVIDIA's marketing suggests. These gaming numbers came off my own test bench: a Ryzen 7 7800X3D system, identical settings on every card, with a capture kept for every run. Everything here was double-checked before it went up.
| Benchmark | Result | 1% Low | VRAM Used | Power | Data |
|---|---|---|---|---|---|
| Alan Wake 2 1440p · High | 79 FPS | 65 | 9.63 GB | 221 W | ✓ Measured |
| Baldur's Gate 3 1440p · Ultra | 158 FPS | 121 | 7.16 GB | 250 W | ✓ Measured |
| Black Desert Online 1440p · Remastered | 158 FPS | 74 | 5.36 GB | 236 W | ✓ Measured |
| Black Myth: Wukong 1440p · Very High | 75 FPS | 59 | 7.69 GB | 209 W | ✓ Measured |
| Call of Duty: MW3 1440p · Extreme | 112 FPS | 59 | 10.45 GB | 216 W | ✓ Measured |
| Clair Obscur: Expedition 33 1440p · Epic | 49 FPS | 43 | 8.03 GB | 204 W | ✓ Measured |
| Counter-Strike 2 1440p · Very High | 234 FPS | 129 | 7.04 GB | 213 W | ✓ Measured |
| Cyberpunk 2077 1440p · Ultra | 103 FPS | 81 | 7 GB | 251 W | ✓ Measured |
| Diablo 4 1440p · Ultra | 131 FPS | 88 | 10.21 GB | 228 W | ✓ Measured |
| Escape from Tarkov 1440p · Ultra | 124 FPS | 90 | 11.51 GB | 211 W | ✓ Measured |
| Marvel Rivals 1440p · Ultra | 69 FPS | 40 | 8.8 GB | 222 W | ✓ Measured |
| Monster Hunter Wilds 1440p · Ultra | 68 FPS | 41 | 9.87 GB | 186 W | ✓ Measured |
| Red Dead Redemption 2 1440p · Favor Quality | 113 FPS | 91 | 6.63 GB | 228 W | ✓ Measured |
| The Last of Us Part II 1440p · Very High | 103 FPS | 79 | 9.27 GB | 221 W | ✓ Measured |
| The Witcher 3 1440p · Ultra | 106 FPS | 82 | 7.03 GB | 250 W | ✓ Measured |
| Architecture | Blackwell |
| CUDA cores | 6,144 |
| VRAM | 12GB GDDR7 |
| Memory bus | 192-bit |
| Memory bandwidth | 672 GB/s |
| Boost clock | 2,512 MHz |
| TDP | 250 W |
| Process | 3nm |
| Interface | PCIe 5.0 x16 |
| Release date | 2025-03-05 |
| Launch MSRP | $549 |
GeForce RTX 5070 averages 112 FPS at 1440p on our suite. These are our own numbers. 15 games, one bench, one card, with the capture published for each result.
100% = this card, Gaming headline metric (Avg FPS). #23 of 82 desktop cards in this vertical.
| GPU | Relative | % | Avg FPS |
|---|---|---|---|
| NVIDIA GeForce RTX 4070 Super | 112% | 125 | |
| AMD Radeon RX 6900 XT | 109% | 122 | |
| AMD Radeon RX 9070 | 108% | 121 | |
| AMD Radeon RX 7800 XT | 105% | 118 | |
| GeForce RTX 5070 | 100% | 112 | |
| NVIDIA GeForce RTX 4070 | 98% | 110 | |
| NVIDIA Titan RTX | 93% | 104 | |
| NVIDIA GeForce RTX 2080 Ti Founders Edition | 89% | 100 | |
| NVIDIA GeForce RTX 3080 | 89% | 100 |
Same card, other workloads: GeForce RTX 5070 AI benchmarks, LLM tokens/sec, image generation & VRAM fit
| Transistors | 31,100 million |
| Die size | 263 mm² |
| Process node | 4 nm |
| Fabricated by | TSMC |
| Transistor density | 118.3 million per mm² |
Denser than 92% of the 746 cards we have silicon data for. Density is the clearest measure of what a process node bought: a card that gained it without growing the die got its speed from the fab rather than the architecture.
Silicon figures from Wikipedia (CC BY-SA 4.0). Benchmarks on this page are our own. Compare every chip.
Whole-job timings, composed from our measured per-model results on this card.
| Workflow | Time | Energy | Basis |
|---|---|---|---|
| Full codebase review | 17.2 min | 45.69 Wh | measured |
Can't run: Product photo shoot (needs FLUX.1 Kontext dev), Photo restoration batch (needs FLUX.1 Kontext dev), Restore and enlarge photos (needs FLUX.1 Kontext dev), Character sheet, 12 poses (needs FLUX.1 dev), Short social clips (needs Qwen3 32B), 60-second AI short film (needs Qwen3 32B), 24-frame storyboard (needs Qwen3 32B), 6-panel comic page (needs Qwen3 32B), Long-form article batch (needs Llama 3.3 70B).
This card is $549 to buy. The cheapest listed rate on Vast.ai is $0.162/hour, but that is the floor: we budget $0.194/hour, a 20% premium, because idle time, storage and unavailable cheap instances all land on the same bill. At that rate buying wins after 2,824 GPU-hours. Below it you are paying for idle silicon.
| How you would use it | GPU-hours a year | Rental cost a year | Time to break even |
|---|---|---|---|
| 2 hours a day, hobby | 730 | $142 | 3.9 years |
| 8 hours a day, working on it | 2,920 | $568 | 11.6 months |
| 24/7, always-on agent | 8,760 | $1,703 | 3.9 months |
At steady usage this card pays for itself inside a normal ownership window. Rental figures include a 20% premium over the cheapest listed rate. Ignores electricity, resale and the fact that a rented card can be a newer one tomorrow.
Cheapest of the RunPod and Vast on-demand rates we see, sampled daily. Spot and interruptible pricing runs lower.