NVIDIA GeForce RTX 4090
Hacker ApprovedRanked 8 of 42 graphics cards by recommenders
Recommended for fast local AI and gaming performance; people cite its speed and value on the used market. Caveat: its VRAM limits large models and it draws lots of power.
Recommended for fast local AI and gaming performance; people cite its speed and value on the used market. Caveat: its VRAM limits large models and it draws lots of power.
49 people gave an opinion
- recommend it
- 38
- criticize it
- 11
- approval share
- 78%
Recommended in 12 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 38 people recommend itNeeds 10 or more, met.
- 3.5 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 12 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Very fast for local models
“RedHatAI/gemma-4-26B-A4B-it-FP8-dynamic on the dual rtx4090 machine under vLLM absolutely slays at token speed.”
Comment from 22 Jul 2026 in “Nvidia DGX Spark as a daily driver” “performance wise it's there at 90-100 t/s on 2x4090, great intelligence”
Comment from 19 Aug 2026 in “Ornith-1.5: From Self-Scaffolding to Self-Improvement” “the RTX 4090 is an entirely different experience (~170 token/sec)”
Comment from 6 Jul 2026 in “AMD Ryzen AI Halo – $4k AI Dev Kit” - Comment from 16 Jun 2026 in “Running local models is good now”
- Comment from 29 Jun 2026 in “Qwen 3.6 27B is the sweet spot for local development”
Strong gaming performance
“I haven't had a single issue with any of these games.”
Comment from 8 Dec 2025 in “Wayland Nvidia” “you can get a long way with a high-end gaming rig”
Comment from 5 Nov 2025 on Hacker News “a decent gaming GPU (RTX 4090 in my case)”
Comment from 16 Jun 2026 in “Running local models is good now” Good value on used market
“Go at ebay and search for RTX 4090 48GBs. There's plenty of them with prices around $3.5k”
Comment from 27 Mar 2026 in “Apple discontinues the Mac Pro with no plans for future hardware” “3090 and 4090 are amazing value for the money on the second hand market”
Comment from 3 Oct 2026 in “From the creator of Redis; run LLM locally with ds4” “Get a second hand 3090/4090 or buy a new Intel Arc Pro B70.”
Comment from 12 Apr 2026 in “Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage” - Comment from 23 Aug 2026 in “I gave Qwen 3.8 27B a reverse-engineering job and it finished in 30 minutes”
Hard to beat for AI
“I don't think there's any reasonable upgrade path right nor or in the near future which would have been better than just buying that.”
Comment from 8 Aug 2026 in “All memory manufactured in 2027 has been sold out” “If you need speed, I think most people wind up with something like a gaming PC with a couple 3090 or 4090s in it.”
Comment from 21 Mar 2026 in “Tinybox- offline AI device 120B parameters” “I think you will be much better with a couple of RTX 5090,4090 or 3090.”
Comment from 23 Dec 2025 in “GLM-4.7: Advancing the Coding Capability”
What people warn about
Each mark is one comment
Limited VRAM for large models
“only has enough vram to load weights and about 6K context”
Comment from 21 Feb 2026 in “With Nvidia's GB10 Superchip, I'm Running Serious AI Models in My Living Room” “rather than having its own walled-off maybe 16 GB VRAM in mainstream gaming GPUs or 24 GB in RTX 4090”
Comment from 22 Jun 2026 on Hacker News High power draw
“Why isn't the RTX 4090 or 5090 half the TDP?”
Comment from 28 Nov 2025 in “Same-day upstream Linux support for Snapdragon 8 Elite Gen 5” “They consume a lot of power”
Comment from 1 Jun 2026 in “Nvidia RTX Spark”
Other graphics cards people recommend
The four most recommended other graphics cards, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 3090 | 116 | 21 | 85% | 12 | 5 yrs14 owners | Approved |
| NVIDIA GeForce RTX 5090 | 102 | 37 | 73% | 12 | 1 yr3 owners | Approved |
| NVIDIA RTX PRO 6000 | 77 | 26 | 75% | 12 | — | Approved |
| Nvidia GPUs | 67 | 67 | 50% | 11 | 6 yrs3 owners | — |
| NVIDIA GeForce RTX 4090This page | 38 | 11 | 78% | 12 | — | Approved |
What people said
40 of the 53 comments that mention it, newest first. Each links to the original comment.
“or RTX 4090 with 48 Gb. The modded cards are better (by price and RAM size)”
“the BAR1 it advertises (32G) is smaller than the VRAM so not all software is able to use it”
“the 4090 would have been a better option, if we'd known what was coming!”
“it turned out it's nice for things like this, but it's hardly in the 99% can't afford domain”
“3090 and 4090 are amazing value for the money on the second hand market”
“With only a RTX 4090, you will reach 30tps”
“I already had an RTX 4090 and I forked ninfer with the obsession to squeeze everything out of this card”
“that thing now is worth its weight in gold”
“handling around 10 concurrent requests at 50 ms TTFA after some config changes”
“performance wise it's there at 90-100 t/s on 2x4090, great intelligence”
“I don't think there's any reasonable upgrade path right nor or in the near future which would have been better than just buying that.”
“RedHatAI/gemma-4-26B-A4B-it-FP8-dynamic on the dual rtx4090 machine under vLLM absolutely slays at token speed.”
“the RTX 4090 is an entirely different experience (~170 token/sec)”
“4090s or 5090s if you can afford it. Way faster.”
“But it gets nowhere close on fp64, where most of the physics simulations and other usual supercomputer tasks shine.”
“it's great but DoA for even quantized LLMs”
“The RTX, unsurprisingly, is all around superior performance wise”
“rather than having its own walled-off maybe 16 GB VRAM in mainstream gaming GPUs or 24 GB in RTX 4090”
“the 4090 performs well with it”
“a decent gaming GPU (RTX 4090 in my case)”
“Works perfectly fine in llama.cpp throwing 70+t/s at me”
“I think it's not really worth it compared to just buying tokens or a coding plan.”
“They consume a lot of power”
“I don't need anything more than a 4090 on Vast.ai”
“the superior performance of a 4090 or a 5090 can be objectively measured”
“it's a pain tweaking the params to run it on a 4090”
“Cyberpunk 2077 struggled even with my 4090, so I had to disable ray tracing and enable DLSS.”
“Get a second hand 3090/4090 or buy a new Intel Arc Pro B70.”
“Go at ebay and search for RTX 4090 48GBs. There's plenty of them with prices around $3.5k”
“If you need speed, I think most people wind up with something like a gaming PC with a couple 3090 or 4090s in it.”
“only has enough vram to load weights and about 6K context”
“they are a legit repair shop being transparent about everything, no one is in the dark they are getting a new product”
“I think you will be much better with a couple of RTX 5090,4090 or 3090.”
“Ada cards beat Blackwell cards in performance per watt”
“I haven't had a single issue with any of these games.”
“two of them (or three 3090/4090s) have the same total memory as your Mac”
“my current system with 64GB of memory and a 4090 is likely to be good for years to come”
“Give me one (or several) 3090/4090/5090 any day of the week over a Mac.”
“Why isn't the RTX 4090 or 5090 half the TDP?”
“you can get a long way with a high-end gaming rig”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.