NVIDIA announced the first desktop GeForce RTX 50-series cards at CES on January 6, 2025, but they did not all go on sale that day. The RTX 5090 and RTX 5080 launched January 30; the RTX 5070 Ti and RTX 5070 followed in February. NVIDIA later added the RTX 5060 Ti and RTX 5060. Official U.S. launch prices ranged from $299 for the RTX 5060 to $1,999 for the RTX 5090.
The cards use NVIDIA’s Blackwell architecture and support DLSS 4, including Multi Frame Generation on RTX 50-series GPUs. Those features can raise displayed frame rates, but NVIDIA’s largest performance claims rely on supported games and AI-generated frames—not simply twice the conventional rendering performance. The prices below are launch MSRPs, not a survey of current retail prices.
RTX 50-series prices and release dates
The January 2025 announcement covered four desktop cards. NVIDIA expanded the desktop lineup with the RTX 5060 family later that spring. The dates and prices below are NVIDIA’s announced U.S. starting prices and initial availability; partner-card prices and present-day stock can differ.
| GPU | Official U.S. starting price | Availability | Memory |
|---|---|---|---|
| GeForce RTX 5090 | $1,999 | January 30, 2025 | 32GB GDDR7 |
| GeForce RTX 5080 | $999 | January 30, 2025 | 16GB GDDR7 |
| GeForce RTX 5070 Ti | $749 | February 2025 | 16GB GDDR7 |
| GeForce RTX 5070 | $549 | February 2025 | 12GB GDDR7 |
| GeForce RTX 5060 Ti 16GB | $429 | April 16, 2025 | 16GB GDDR7 |
| GeForce RTX 5060 Ti 8GB | $379 | April 16, 2025 | 8GB GDDR7 |
| GeForce RTX 5060 | $299 | May 19, 2025 | 8GB GDDR7 |
These are launch prices, not guaranteed current prices. The least expensive card meeting NVIDIA’s starting-price description may be a basic partner design; factory-overclocked or premium-cooled versions can cost more. For current stock and pricing, check retailer listings rather than treating MSRP as a live quote. NVIDIA’s RTX 50-series product finder helps identify partner models, but does not guarantee their availability or price.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
What NVIDIA announced at CES 2025
On January 6, 2025, NVIDIA introduced the Blackwell-based RTX 5090, RTX 5080, RTX 5070 Ti, and RTX 5070, alongside RTX 50-series laptop GPUs and new graphics and AI features. The announcement included DLSS 4, Multi Frame Generation, Reflex 2, RTX Neural Shaders, and RTX Mega Geometry. The RTX 5090 and RTX 5080 were the first desktop cards scheduled to reach retail, on January 30—not all four models at once. NVIDIA’s CES overview and RTX 50-series announcement detail the initial reveal.
NVIDIA also announced mobile versions. Its initial timing put RTX 5090, RTX 5080, and RTX 5070 Ti laptops in March 2025, with RTX 5070 laptops following in April. A laptop GPU with the same model number as a desktop card is not a direct performance equivalent: laptop power limits, cooling, and system design affect results.
RTX 50-series desktop specifications
The table compares the principal desktop models using NVIDIA’s published specifications. AI TOPS is a processor-performance figure, not a direct measure of game frame rate. Partner boards can vary in clock speeds, dimensions, cooling, connectors, and power limits. NVIDIA’s comparison page provides its current product specifications.
| GPU | CUDA cores | Memory | Memory bus | NVIDIA AI TOPS | General positioning |
|---|---|---|---|---|---|
| RTX 5090 | 21,760 | 32GB GDDR7 | 512-bit | 3,352 | Enthusiast 4K, creator, and AI workloads |
| RTX 5080 | 10,752 | 16GB GDDR7 | 256-bit | 1,801 | High-end 4K |
| RTX 5070 Ti | 8,960 | 16GB GDDR7 | 256-bit | 1,406 | High-refresh 1440p and 4K |
| RTX 5070 | 6,144 | 12GB GDDR7 | 192-bit | 988 | 1440p |
| RTX 5060 Ti | 4,608 | 8GB or 16GB GDDR7 | 128-bit | 759 | Mainstream 1080p and 1440p |
| RTX 5060 | 3,840 | 8GB GDDR7 | 128-bit | 614 | Mainstream 1080p |
What Blackwell and DLSS 4 change
Blackwell combines updated shader, ray-tracing, and Tensor hardware with GDDR7 memory and a stronger emphasis on neural rendering. NVIDIA’s design adds AI-assisted techniques to the conventional graphics pipeline rather than relying only on the GPU drawing every displayed frame directly. The company also describes improvements for video and creator workflows. Details of the architecture are available in NVIDIA’s Blackwell architecture reference.
DLSS 4 is a collection of features, not one universal quality or performance setting. Depending on the game, it can include Super Resolution, DLAA, Ray Reconstruction, Frame Generation, and Multi Frame Generation. Availability varies by game and application. Multi Frame Generation is for RTX 50-series GPUs and can synthesize up to three additional frames for each traditionally rendered frame, according to NVIDIA’s RTX 5060 launch announcement.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Rendered frames versus generated frames
A conventionally rendered frame is produced by the game’s rendering pipeline. A generated frame is synthesized from image and motion information between rendered frames. A monitor or FPS overlay may count both in the displayed frame rate, so that number can rise without an equal increase in conventionally rendered frames.
Generated frames can make motion appear smoother, but they do not automatically provide the same responsiveness as a higher base frame rate. NVIDIA Reflex is intended to reduce system latency, and Reflex 2 was part of the CES announcement. The real result depends on the game, settings, support, and underlying performance; players sensitive to latency should pay attention to responsiveness, not only the displayed FPS.
How to read NVIDIA’s performance claims
NVIDIA promoted up-to-two-times performance comparisons for several RTX 50 models, including RTX 5090 versus RTX 4090, RTX 5080 versus RTX 4080, and RTX 5070 versus RTX 4070. Those are company claims for selected scenarios, not a blanket statement that every game runs twice as fast. Some prominent comparisons depend on DLSS 4 and Multi Frame Generation, so they should not be read as equivalent gains in native rendering.
NVIDIA also positioned the RTX 5070 as offering RTX 4090-like performance. That comparison is workload- and feature-dependent; it does not mean the two cards are universally equivalent in conventional rendering, latency, or every game. To assess a particular claim, look for the named game, resolution, preset, ray-tracing settings, DLSS mode, and whether Multi Frame Generation is on. The announcement’s claims are described in NVIDIA’s launch coverage. Independent game-by-game testing is needed to judge performance outside those selected demonstrations.
What each desktop model is suited to
RTX 5090: maximum capacity and performance
With 32GB of GDDR7, the RTX 5090 is the flagship for buyers seeking maximum performance, substantial graphics memory, or demanding 4K, rendering, and AI workloads. Its $1,999 launch MSRP puts it far above the RTX 5080; that price is difficult to justify for ordinary 1080p gaming. Its size and power demands also make system compatibility especially important.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
RTX 5080: high-end 4K
The RTX 5080 is positioned for high-end 4K gaming and has 16GB of memory. Buyers with demanding games, high-resolution textures, or creator projects should consider whether that capacity fits their workloads, as well as comparing actual card prices and independent results.
RTX 5070 Ti: 16GB at a lower launch tier
The RTX 5070 Ti also has 16GB, while its launch MSRP was $250 below the RTX 5080’s. That makes it a plausible enthusiast option for high-refresh 1440p or 4K, but value depends on real retail prices and performance in the games and applications you use.
Recommended Free Tools
RTX 5070: a 1440p option with 12GB
The RTX 5070 targets 1440p and includes 12GB of GDDR7. VRAM use varies with resolution, textures, ray tracing, mods, and game engine; the capacity could become a constraint in some demanding settings, but no single minimum applies to every future game.
RTX 5060 Ti: choose the memory version deliberately
The RTX 5060 Ti comes in 8GB and 16GB versions. They share a model name and CUDA-core count, but the extra memory can matter for high texture settings, modded games, creator work, and longer ownership. For those uses, compare the 16GB model’s price against the 8GB version rather than treating them as interchangeable.
RTX 5060: entry-level desktop RTX 50
The RTX 5060 has 8GB of GDDR7 and is aimed at mainstream 1080p gaming. It is most compelling for a buyer whose games and preferred settings fit that memory capacity and who can benefit from supported DLSS features.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5080
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Choosing a card for your games and system
Start with resolution, refresh rate, and game support
- 1080p: The RTX 5060 or an RTX 5060 Ti may suit a mainstream system, depending on the game, desired settings, and price. Competitive players should prioritize a consistently strong base frame rate and low latency.
- 1440p: The RTX 5070 is a natural starting point; consider the RTX 5070 Ti if you want more headroom or value its 16GB memory. Check game-specific tests rather than assuming one model will perform the same across every title.
- 4K: The RTX 5080 and RTX 5090 are the higher-end options in this lineup, while the 5070 Ti may suit some users depending on settings and games. Memory needs, ray tracing, and DLSS support influence the trade-off.
- Games without DLSS support: Base rendering performance matters more when a game does not support the features behind NVIDIA’s largest frame-generation claims.
Creator and AI workloads
For editing, 3D work, livestreaming, or AI software, assess the exact application’s CUDA, encoding, ray-tracing, and memory requirements rather than using gaming FPS as a proxy. NVIDIA’s Studio information covers creator workflows. GeForce is not a substitute for a workstation card when a job specifically requires certified drivers, ECC memory, enterprise support, or workstation-class reliability.
Upgrading from an older RTX card
For RTX 40-series owners, compare independent results in your games with the settings you actually use, including native rendering if you do not use frame generation. If your current card already meets your monitor’s needs, a new feature alone may not justify an upgrade. RTX 30-series owners may have a stronger case if a newer card materially improves their target resolution, ray tracing, or supported DLSS workflow, but the right choice still depends on price and measured performance.
Founders Edition, partner cards, and real prices
NVIDIA offered limited Founders Edition models of the RTX 5090, RTX 5080, and RTX 5070. Board partners also sold stock-clocked and factory-overclocked versions of the original four desktop GPUs. Their cards can differ in length, thickness, cooling, noise, factory clock, power limit, connector placement, and warranty. The starting MSRP does not describe every version’s price or guarantee that a card is available at that price.
Since launch prices are historical, compare current listings for the exact card, including tax and shipping, before buying. A higher-priced factory-overclocked card may offer only a modest gain; weigh its cooling, noise, warranty, and price against a less expensive model. NVIDIA’s specification comparison and product finder are useful for checking model details. A contemporary price-watch source is PC Gamer’s graphics-card price watch; it is not an official NVIDIA price list.
Quick Recap
Compatibility checklist before buying
- Case clearance: Check the exact card’s length and thickness against the case and nearby components. Large triple-fan designs may block adjacent expansion slots.
- Power supply: Verify the specific card maker’s recommended wattage and required power connector. Partner models may have different requirements; there is no single PSU recommendation for every RTX 50 card.
- Power-cable routing: Confirm the connector arrangement works with your case and that the cable can be routed without pressing against a side panel.
- Cooling and airflow: Check that the case can move enough air for the card, especially for a flagship model or a compact build.
- Monitor: Match the GPU to your display’s resolution and refresh rate; paying for high-end 4K performance is of little use on a modest 1080p display.
- Features and drivers: Confirm that your games support the DLSS features you intend to use and install current drivers. NVIDIA provides its NVIDIA App for driver and GPU-management features.
- Purchase terms: Check the exact model’s warranty, return policy, and current price before committing.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
