Skip to content
Local AI Wizard
Your machine
Theme

RTX 4090 vs RTX 5090

RTX 4090 · 24 GB VRAM and RTX 5090 · 32 GB VRAM side by side for local AI.

Facts from both hardware pages

RTX 4090 vs RTX 5090: memory, bandwidth, models and prices
SpecificationRTX 4090RTX 5090
Memory24 GB VRAMDedicated · system RAM does not add · source32 GB VRAMDedicated · system RAM does not add · source
Memory bandwidth1,008 GB/sManufacturer spec · not model speed · source1,792 GB/sManufacturer spec · not model speed · source
Text · run well129+7 tight fit · 25 run slowly · within 24 GB158+2 tight fit · 25 run slowly · within 32 GB
Vision · run well126+7 tight fit · within 24 GB140+2 tight fit · within 32 GB
Image · within memory53within 24 GB · documented paths and minimums64within 32 GB · documented paths and minimums
Video · within memory30within 24 GB · documented paths and minimums34within 32 GB · documented paths and minimums
Audio · within memory128within 24 GB · documented paths and minimums132within 32 GB · documented paths and minimums
Embeddings · within memory90within 24 GB · 8-bit estimate90within 32 GB · 8-bit estimate
Published speed runs1313 on this machine1716 on this machine · 1 naming only the chip
Launch price (MSRP)USD 1,599US · dated 2022-10-12 · not a tracked offer · sourceUSD 1,999US · dated 2025-01-30 · not a tracked offer · source
Where to buyNo manufacturer store link on fileNVIDIA Marketplacethis configuration · US · recorded 2026-10-18 · store link

Counts use the same memory estimate as each hardware page: text and vision run well within the fast memory (tight fits counted apart); image, video and audio use documented paths and publisher minimums. Formulas

Top text models each runs well

The first 5 rows of each hardware page's Text list, in its order: LiveBench score, then reviewed models, then Hub downloads. Memory is the planning estimate on that machine.

RTX 4090 vs RTX 5090: top text models that run well
RankRTX 4090RTX 5090
#1Smaug MiniAbacus AI · LiveBench 76.9 · ~23 GBSmaug MiniAbacus AI · LiveBench 76.9 · ~23 GB
#2Qwen3.8 27BAlibaba · LiveBench 75.3 · ~23 GBQwen3.8 27BAlibaba · LiveBench 75.3 · ~23 GB
#3Qwen3.6 27BAlibaba · LiveBench 64 · ~23 GBQwen3.6 27BAlibaba · LiveBench 64 · ~23 GB
#4Qwen3 14BAlibaba · ~14 GBQwen3 14BAlibaba · ~14 GB
#5Gemma 4 26B A4BGoogle · ~21 GBGemma 4 31BGoogle · ~26 GB

Measured speed

Published output tokens per second, as measured. A speed is compared only when the same checkpoint, quantization, runtime and lengths ran on both machines themselves; same-chip references are not compared.

RTX 4090 vs RTX 5090: the same published test on both
Same testRTX 4090RTX 5090
gpt-oss 20B Q4_K_M · llama.cpp CUDA277.95 ± 1.502026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0393.3 ± 2.622026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
gpt-oss 20B Q8_0 · llama.cpp CUDA258.4 ± 1.922026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0374.72 ± 2.732026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Qwen3-Coder 30B-A3B Q4_K_M · llama.cpp CUDA265.96 ± 2.532026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0346.27 ± 3.492026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Llama 2 7B Q4_0 · CUDA188.96 ± 0.05Not stated by source · llama.cpp CUDA scoreboard · community result300.40 ± 0.28Not stated by source · llama.cpp CUDA scoreboard · community result
gpt-oss-20b MXFP4 · llama.cpp CUDA221.95 ± 6.34Not stated by source · llama.cpp gpt-oss guide · maintainer result282.51 ± 0.44Not stated by source · llama.cpp gpt-oss guide · maintainer result
Gemma 4 26B-A4B UD-Q4_K_M · llama.cpp CUDA169.27 ± 1.612026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0231.41 ± 1.952026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Qwen3.5 9B Q4_K_M · llama.cpp CUDA141.55 ± 0.582026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0220.46 ± 1.182026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Qwen3.5 9B Q8_0 · llama.cpp CUDA95.08 ± 0.132026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0156.01 ± 0.802026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Gemma 4 12B Q4_K_M · llama.cpp CUDA101.95 ± 0.082026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0152.43 ± 0.682026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Gemma 4 12B Q8_0 · llama.cpp CUDA64.22 ± 0.092026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0103.67 ± 0.382026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Mistral Small 3.2 24B Q4_K_M · llama.cpp CUDA61.56 ± 0.072026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0101.14 ± 0.242026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Qwen3.6 27B Q4_K_M · llama.cpp CUDA48.38 ± 0.042026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.077.61 ± 0.382026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
Gemma 4 31B Q4_K_M · llama.cpp CUDA44.42 ± 0.042026-10-16 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.071.3 ± 0.232026-10-17 · Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
All published measurements · RTX 409013 on this machine
Published measurements on this hardware
gpt-oss 20B Q4_K_Munsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)277.95 ± 1.50Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3-Coder 30B-A3B Q4_K_Munsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Q…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)265.96 ± 2.53Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
gpt-oss 20B Q8_0unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.g…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)258.4 ± 1.92Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
gpt-oss-20b MXFP4gpt-oss-20b-mxfp4.gguf (ggml-org/gpt-oss-20…NVIDIA GeForce RTX 4090 · 24 GB221.95 ± 6.34MXFP4 (native gpt-oss weights)
0 / 128 · Single llama-bench process;…
llama.cpp CUDA
llama.cpp commit a094f38 (6210)
Flash Attention enabled
llama.cpp gpt-oss guide · maintainer result
2026-10-20
Llama 2 7B Q4_0TheBloke/Llama-2-7B-GGUF · llama-2-7b.Q4_0.…NVIDIA GeForce RTX 4090 · 24 GB GDDR6X188.96 ± 0.05Q4_0
0 / 128 · Not reported by llama-bench
CUDA
llama.cpp commit 2241453
Flash Attention enabled
llama.cpp CUDA scoreboard · community result
2026-09-25
Gemma 4 26B-A4B UD-Q4_K_Munsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)169.27 ± 1.61UD-Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.5 9B Q4_K_Munsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.g…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)141.55 ± 0.58Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 12B Q4_K_Munsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)101.95 ± 0.08Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.5 9B Q8_0unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.ggufNVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)95.08 ± 0.13Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 12B Q8_0unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)64.22 ± 0.09Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Mistral Small 3.2 24B Q4_K_Munsloth/Mistral-Small-3.2-24B-Instruct-2506…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)61.56 ± 0.07Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.6 27B Q4_K_Munsloth/Qwen3.6-27B-GGUF/Qwen3.6-27B-Q4_K_M…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)48.38 ± 0.04Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 31B Q4_K_Munsloth/gemma-4-31B-it-GGUF/gemma-4-31B-it-…NVIDIA GeForce RTX 4090 · 24 GB (the card's only memory size; rented host)44.42 ± 0.04Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
All published measurements · RTX 509016 on this machine · 1 naming only the chip
Published measurements on this hardware
gpt-oss 20B Q4_K_Munsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)393.3 ± 2.62Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
gpt-oss 20B Q8_0unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q8_0.g…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)374.72 ± 2.73Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3-Coder 30B-A3B Q4_K_Munsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/Q…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)346.27 ± 3.49Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Llama 2 7B Q4_0TheBloke/Llama-2-7B-GGUF · llama-2-7b.Q4_0.…NVIDIA GeForce RTX 5090 · 32 GB GDDR7300.40 ± 0.28Q4_0
0 / 128 · Single llama-bench process;…
CUDA
llama.cpp commit 8cf6b42
Flash Attention enabled
llama.cpp CUDA scoreboard · community result
2026-09-26
gpt-oss-20b MXFP4gpt-oss-20b-mxfp4.gguf (ggml-org/gpt-oss-20…NVIDIA GeForce RTX 5090 · 32 GB282.51 ± 0.44MXFP4 (native gpt-oss weights)
0 / 128 · Single llama-bench process;…
llama.cpp CUDA
llama.cpp commit a094f38 (6210)
Flash Attention enabled
llama.cpp gpt-oss guide · maintainer result
2026-10-20
Llama 2 7B Q4_0Llama 2 7B Q4_0 (scoreboard model; the scor…Nvidia RTX 5090 · 32 GB (memory size not stated by the source; this chip is sold with one memory size) · coopmat2273.68 ± 0.52Q4_0
0 / 128 · Single llama-bench process;…
Vulkan
llama.cpp commit ca71fb9
Flash Attention enabled
llama.cpp Vulkan scoreboard · community result
2026-10-25
Gemma 4 26B-A4B UD-Q4_K_Munsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)231.41 ± 1.95UD-Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.5 9B Q4_K_Munsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q4_K_M.g…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)220.46 ± 1.18Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 26B-A4B Q8_0unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)217.48 ± 1.82Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.5 9B Q8_0unsloth/Qwen3.5-9B-GGUF/Qwen3.5-9B-Q8_0.ggufNVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)156.01 ± 0.80Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 12B Q4_K_Munsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)152.43 ± 0.68Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 12B Q8_0unsloth/gemma-4-12b-it-GGUF/gemma-4-12b-it-…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)103.67 ± 0.38Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Mistral Small 3.2 24B Q4_K_Munsloth/Mistral-Small-3.2-24B-Instruct-2506…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)101.14 ± 0.24Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.6 27B Q4_K_Munsloth/Qwen3.6-27B-GGUF/Qwen3.6-27B-Q4_K_M…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)77.61 ± 0.38Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Gemma 4 31B Q4_K_Munsloth/gemma-4-31B-it-GGUF/gemma-4-31B-it-…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)71.3 ± 0.23Q4_K_M
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Mistral Small 3.2 24B Q8_0unsloth/Mistral-Small-3.2-24B-Instruct-2506…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)63.85 ± 0.10Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25
Qwen3.6 27B Q8_0unsloth/Qwen3.6-27B-GGUF/Qwen3.6-27B-Q8_0.g…NVIDIA GeForce RTX 5090 · 32 GB (the card's only memory size; rented host)52.36 ± 0.22Q8_0
0 / 128 · Single llama-bench process
llama.cpp CUDA
llama.cpp commit 3e4dfa8f3 (b11658)
Flash Attention enabled
Data: Local LLM Advisor (localllmadvisor.com) · CC BY 4.0
2026-10-25

Verdict

RTX 4090 has 24 GB VRAM and RTX 5090 has 32 GB VRAM: 8 GB more on RTX 5090. By the memory estimate, 129 text models run well on RTX 4090 and 158 on RTX 5090. RTX 5090 lists 1,792 GB/s of memory bandwidth, 1.8× the 1,008 GB/s of RTX 4090 (a manufacturer spec, not a measured speed). In the same published test (gpt-oss 20B Q4_K_M, llama.cpp CUDA), RTX 4090 measured 278 and RTX 5090 393.3 output tokens per second. Launch prices on file: RTX 4090 USD 1,599 (2022-10-12) and RTX 5090 USD 1,999 (2025-01-30). RTX 5090 leads on both facts here: more text models run well and it measured faster in the shared test.

Limits and dates

Model counts are memory estimates, not speed measurements or guaranteed fits; verify runtime and checkpoint support. Bandwidth is the manufacturer's figure. Prices are launch or list prices with their dates, not current offers.

Page updated · Model catalog checked · LiveBench 2026-06-25