The RTX 50-series is on shelves, which means the 40-series is now the smart-money tier. The 4080 and 4090 went from "halo product flexes" to "the cards used builders actually buy when they want flagship-class performance without paying flagship-class taxes." And right now, on the used and clearance market, that gap between them is where the real conversation lives.

The 4090 is still the faster card. Nobody is arguing that. The question is whether it's faster enough to justify a price tag that is, depending on where you look, $500 to $700 above what an RTX 4080 Super costs. For most people, the answer is no. For a small slice of buyers, the answer is absolutely yes. This is the breakdown.

RTX 4080 vs RTX 4090 comparison infographic showing specs, VRAM, power draw, and pricing

The Specs That Actually Matter

The RTX 4080 ships with 9,728 CUDA cores, 16GB of GDDR6X on a 256-bit bus, and a 320W board power target. The 4080 Super (functionally the same card with a 5% core bump and a $200 MSRP cut) sits at the same TBP. New retail is around $999–$1,199 when you can find one. Used 4080 Supers on eBay run $700–$880.

The RTX 4090 packs 16,384 CUDA cores — that's 68% more shader hardware — paired with 24GB of GDDR6X on a 384-bit bus. Power draw jumps to 450W. New 4090s, where they exist, run $1,599 MSRP and up (street pricing has been brutal). Used 4090s have stayed weirdly expensive — $1,400–$1,700 — because the 5090 is even more expensive and the used 4090 became the "I need 24GB of VRAM but won't pay 5090 money" card.

That VRAM number is the single most important spec on this page. 16GB vs 24GB doesn't matter for any current game at 4K. It matters enormously if you're touching local AI, large-scale 3D work, or any workload where your model has to fit in memory or it doesn't run at all. If you're a gamer, this spec is irrelevant for the next 2–3 years. If you're not just a gamer, this spec is the reason the 4090 still costs what it costs.

RTX 4080 vs RTX 4090 — specs at a glance
Street pricing as of April 2026
RTX 4080 / Super
NVIDIA · Ada Lovelace
CUDA cores9,728
VRAM16 GB GDDR6X
Memory bus256-bit
Bandwidth717 GB/s
TDP320W
PCIe4.0 x16
Used street~$800
VS
RTX 4090
NVIDIA · Ada Lovelace
CUDA cores16,384
VRAM24 GB GDDR6X
Memory bus384-bit
Bandwidth1,008 GB/s
TDP450W
PCIe4.0 x16
Used street~$1,500

Performance: The Numbers Are What They Are

Aggregated across a 20-game benchmark suite, the RTX 4090 is roughly 27–28% faster than the RTX 4080 at 4K. That number is consistent whether you look at Hardware Times, Technical City, or NanoReview. It's not a controversial figure.

What changes is which games get hit hardest by that gap:

  • 4K, max settings, no upscaling: 4090 is 25–40% faster, depending on title. Cyberpunk 2077 with path tracing on, Alan Wake 2, MSFS 2024 — all show the 4090 stretching its legs. The 4080 is hitting 100 FPS, the 4090 is hitting 130.
  • 1440p, max settings: The gap collapses to 12–18%. The 4080 is already so fast at 1440p that the 4090's extra silicon spends a chunk of its time waiting on the CPU. If you have a high-refresh 1440p monitor, the 4090's premium is mostly invisible.
  • 1080p: Don't buy either of these for 1080p. You're CPU-bound by frame 2.
  • Ray tracing, 4K: Biggest 4090 wins. The 24GB VRAM headroom and the 191 RT cores (vs the 4080's 113) mean the 4090 can sustain RT at 4K in titles where the 4080 has to drop settings or lean on DLSS Performance.

The math nobody likes: at 4K with DLSS Quality on, the 4080 is fast enough for everything currently shipped. The 4090 is fast enough for everything currently shipped, with a buffer for whatever lands in the next 24 months. That buffer is the entire pitch.

RTX 4080 vs RTX 4090 benchmark FPS chart across 6 games at 4K resolution showing 25-40% performance gap

Cost Per Frame Is the Most Honest Metric

Hardware Times' 4K benchmark suite worked out to 0.089 FPS per dollar for the 4080 Super (at $999) and 0.066 FPS per dollar for the 4090 (at $1,699). The 4080 Super is roughly 35% better value per frame.

Used market shifts the math but doesn't flip it. Used 4080 Super at $800 lands around 0.111 FPS/$. Used 4090 at $1,500 lands around 0.075 FPS/$. The 4090's premium narrows on the used market, but it doesn't disappear.

Power costs add insult. Over a 3-year ownership window with 4 hours/day of gaming load at average US electricity rates, the 4090 will burn roughly $130–$170 more in electricity than the 4080. That's not life-changing money, but it's a meaningful chunk of the price gap if you measure it.

RTX 4080 vs RTX 4090 cost per frame analysis showing 4080 Super delivers 35% better value per frame

The Verdict

Buy the RTX 4080 (or 4080 Super) if:

  • You game at 1440p, even high-refresh 1440p — the 4090 is wasted on you
  • You game at 4K and you're fine with DLSS Quality being part of your playbook
  • You don't run local AI workloads, large-scale 3D, or 8K video timelines
  • You want lower temps, less noise, and a power budget under 700W PSU territory
  • You found a clean used 4080 Super under $800

Buy the RTX 4090 if:

  • You run local LLMs, Stable Diffusion at scale, or Blender renders that need 24GB
  • You game at native 4K, max settings, no upscaling, and you refuse to touch DLSS
  • You're a content creator where render time directly trades against billable hours
  • You already own a 1000W+ PSU and a case that can dissipate 450W comfortably
  • You want a card that won't feel slow for 3+ years and the price gap doesn't change your life

Everyone else: the 4080 is the answer. The 4090 is faster, but it's not 60% more card for 60% more money. It's about 28% more frames for double the price, and that math only works if VRAM, AI workloads, or "I need the best, period" is part of your buying criteria.

Both cards aged into a sweet spot. The 4080 Super is the smart-money flagship. The 4090 is the prosumer card that happens to also game. Pick the one that matches your actual use case, not your wishlist.