readygpu

NVIDIA GeForce RTX 4090

24 GB · 1,008 GB/s · 450 W

In short

The NVIDIA GeForce RTX 4090 has 24 GB of GDDR6X memory with 1,008 GB/s of bandwidth, which sets how fast it generates text. At 32k context it runs 48 of the 113 open models we track fully in memory; the most capable is Qwen 3.8 27B at Q5_K_M, at about 27 tok/s (estimated). The best new price for the NVIDIA GeForce RTX 4090 in Germany is 3.955 € at eBay; used, it starts at 2.800 € (asking price).

Updated . Speeds are estimates; how they are worked out.

Prices

Every retailer we track in Germany, cheapest first. Buy links may earn us a commission.

RetailerConditionPriceStockCheckedBuy
eBayMSI GeForce RTX 4090 Gaming X Trio 24Gused, asking price2.800 €In stock1 h agoBuy
eBayMSI GeForce RTX 4090 Ventus 3X 24 GB GDDR6Xused, asking price2.830 €In stock1 h agoBuy
eBayPalit RTX 4090 24GB GDDR6X VRAM GameRock OmniBlackused, asking price2.850 €In stock1 h agoBuy
eBayGIGABYTE GeForce RTX 4090 AERO OC 24G Grafikkarteused, asking price2.849 €+ 15 € deliveryIn stock1 h agoBuy
eBayPalit GeForce RTX 4090 GameRock OC 24GB GDDR6X Grafikkarteused, asking price2.850 €+ 15 € deliveryIn stock1 h agoBuy
eBayMSI GeForce RTX 4090 GAMING X TRIO 24GB GDDR6Xused, asking price2.990 €In stock1 h agoBuy
eBayASUS ROG Strix GeForce RTX 4090 OC White 24GB Grafikkarteused, asking price3.000 €In stock1 h agoBuy
eBayPalit GeForce RTX 4090 GameRock OC 24GB GDDR6X Grafikkarteused, asking price3.000 €In stock1 h agoBuy
eBayGigabyte Geforce RTX 4090 Windforce v2 24G Grafikkarteused, asking price3.000 €In stock1 h agoBuy
eBayMSI GeForce RTX 4090 SUPRIM x Liquid 24g NVIDIA 24GB GDDR6X Grafikkarteused, asking price3.000 €+ 8 € deliveryIn stock5 h agoBuy
eBayMSI GeForce RTX 4090 SUPRIM LIQUID 24GB GDDR6X Grafikkarteused, asking price3.150 €In stock1 h agoBuy
eBayPalit GeForce RTX 4090 GameRock OC 24GB GDDR6X Grafikkarteused, asking price3.150 €+ 8 € deliveryIn stock1 h agoBuy
eBayNVIDIA GeForce RTX 4090 Founders Edition 24GB Grafikkarte inkl. Stromadapterused, asking price3.200 €+ 28 € deliveryIn stock1 h agoBuy
eBayASUS TUF Gaming GeForce RTX 4090 24GB GDDR6X GPU Grafikkarteused, asking price3.299 €In stock1 h agoBuy
eBay24GB ASUS TUF Gaming Nvidia RTX O24G OG OC Edition GDDR6X 4090 GrafikkarteNew3.955 €+ 15 € deliveryIn stock1 h agoBuy

Used on eBay: median asking price 3.200 €, most between 2.955 € and 3.463 € (24 listings, 2026-10-15).

Price history

Lowest price shown each day over the last 90 days.

Not enough price history yet.

What it runs

Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.

ModelQuantFitsMemoryMax contextDecode @32kPrefillGood for
Qwen 3.8 27B

Alibaba Qwen

Q5_K_Mfits22.5 GB44k78k with q8 cache

27tok/sest.

1198 tok/schat, coding
Qwen 3.6 27B

Alibaba Qwen

Q5_K_Mfits21.2 GB64k112k with q8 cache

29tok/sest.

1198 tok/schat, coding
Gemma 4 31B

Google

Q4_K_Mfits21.2 GB57k110k with q8 cache

29tok/sest.

1067 tok/schat, coding
Qwen 3.6 35B-A3B

Alibaba Qwen

Q4_K_Mfits22.3 GB74k116k with q8 cache

145tok/sest.

8550 tok/schat, coding, agents, long documents
Qwen 3.5 27B

Alibaba Qwen

Q5_K_Mfits21.3 GB62k110k with q8 cache

29tok/sest.

1198 tok/schat, coding
Gemma 4 26B-A4B

Google

UD-Q6_Kfits23.3 GB36k62k with q8 cache

84tok/sest.

7606 tok/schat, coding, agents, long documents
Qwen 3.5 35B-A3B

Alibaba Qwen

Q4_K_Mfits22.1 GB84k131k with q8 cache

146tok/sest.

8550 tok/schat, coding, agents, long documents
GLM 4.7 Flash

Z.ai

Q5_K_Mfits22.5 GB48k82k with q8 cache

70tok/sest.

4464 tok/schat, coding, agents
Devstral Small 2 24B

Mistral AI

Q5_K_Mfits21.5 GB43k79k with q8 cache

28tok/sest.

1264 tok/schat, coding
gpt-oss 20B

OpenAI

MXFP4fits12.9 GB128k

116tok/sest.

6383 tok/schat, coding, agents, long documents
Qwen3 Coder 30B-A3B

Alibaba Qwen

Q4_K_Mfits21.2 GB54k97k with q8 cache

74tok/sest.

3932 tok/schat, coding, agents
Gemma 4 12B

Google

Q8_0fits13.5 GB256k

46tok/sest.

2777 tok/schat, coding, long documents
Qwen 3.5 9B

Alibaba Qwen

Q8_0fits10.8 GB256k

59tok/sest.

3461 tok/schat, coding, long documents
Mistral Small 3.2 24B

Mistral AI

Q5_K_Mfits21.5 GB43k79k with q8 cache

28tok/sest.

1264 tok/schat, coding
Gemma 3 27B

Google

Q5_K_Mfits21.7 GB51k95k with q8 cache

28tok/sest.

1271 tok/schat, coding
Ministral 3 14B

Mistral AI

Q8_0fits19.3 GB57k104k with q8 cache

32tok/sest.

1922 tok/schat, coding
Qwen3 14B

Alibaba Qwen

Q8_0fits20.5 GB40k

30tok/sest.

1728 tok/schat, coding
DeepSeek R1 0528 Qwen3 8B

DeepSeek

Q8_0fits13.5 GB99k128k with q8 cache

46tok/sest.

2851 tok/schat, coding
Qwen3 8B

Alibaba Qwen

Q8_0fits13.5 GB40k

46tok/sest.

2851 tok/schat, coding
Phi-4 14B

Microsoft

Q8_0fits21.6 GB16k

28tok/sest.

1737 tok/schat, coding
Qwen 3.5 4B

Alibaba Qwen

Q8_0fits6.1 GB256k

112tok/sest.

6475 tok/schat, coding, long documents
Ministral 3 8B

Mistral AI

Q8_0fits13.5 GB103k187k with q8 cache

46tok/sest.

2754 tok/schat, coding
Granite 4.2 8B

IBM

Q8_0fits14.6 GB86k128k with q8 cache

43tok/sest.

2622 tok/schat, coding, agents
Gemma 4 E4B

Google

Q8_0fits9.4 GB128k

68tok/sest.

4348 tok/schat, coding, long documents
LFM2.5 8B-A1B

Liquid AI

Q8_0fits10.8 GB125k

140tok/sest.

14221 tok/schat, coding, long documents
Llama 3.1 8B

Meta

Q6_Kfits11.0 GB126k128k with q8 cache

57tok/sest.

3012 tok/schat, coding, long documents
Qwen2.5 0.5B

Alibaba Qwen

Q8_0fits1.8 GB32k

669tok/sest.

30972 tok/schat, coding, long documents
Qwen3.5 0.8B

Alibaba Qwen

Q8_0fits2.0 GB256k

507tok/sest.

29093 tok/schat, coding, agents, long documents
Qwen3.5 2B

Alibaba Qwen

Q5_K_Mfits2.6 GB256k

337tok/sest.

13870 tok/schat, coding, agents, long documents
DeepSeek R1 Distill Qwen 1.5B

DeepSeek

Q8_0fits3.5 GB128k

221tok/sest.

11651 tok/schat, coding, long documents
MiniCPM5 2B

openbmb

Q8_0fits4.7 GB128k

153tok/sest.

6958 tok/schat, coding, agents, long documents
LFM2.5 2.6B

Liquid AI

Q8_0fits5.4 GB128k

128tok/sest.

7881 tok/schat, coding, agents, long documents
LFM2.5 VL 3B

Liquid AI

Q8_0fits5.4 GB32k

128tok/sest.

7227 tok/schat, coding, agents, long documents
Mellum2.1 12B A2.5B Thinking

JetBrains

Q8_0fits13.4 GB128k

118tok/sest.

10268 tok/schat, coding, agents, long documents
granite vision 4.1 4b

IBM

Q8_0fits6.8 GB128k

99tok/sest.

5049 tok/schat, coding, long documents
granite 4.2 3b

IBM

Q8_0fits7.0 GB128k

95tok/sest.

5292 tok/schat, coding, agents, long documents
Kimi VL A3B

Moonshot AI

mlx-8bitfits18.4 GB128k

88tok/sest.

7715 tok/schat, coding, agents, long documents
LFM2.5 24B A2B

Liquid AI

Q6_Kfits23.1 GB35k63k with q8 cache

84tok/sest.

7925 tok/schat, coding, agents
DeepSeek R1 Distill Qwen 7B

DeepSeek

Q8_0fits10.2 GB128k

63tok/sest.

3405 tok/schat, coding, long documents
GLM 4.6V Flash

Z.ai

Q8_0fits11.5 GB128k

55tok/sest.

2371 tok/schat, coding, agents, long documents
DeepSeek R1 Distill Llama 8B

DeepSeek

Q8_0fits12.8 GB112k128k with q8 cache

49tok/sest.

3012 tok/schat, coding
GLM 4.1V 9B Thinking

Z.ai

mlx-8bitfits13.1 GB64k

48tok/sest.

2371 tok/schat, coding, long documents
Ministral 3 8B Reasoning 2512

Mistral AI

Q8_0fits13.5 GB103k187k with q8 cache

46tok/sest.

2754 tok/schat, coding, agents
DeepSeek R1 Distill Qwen 14B

DeepSeek

Q8_0fits21.5 GB41k76k with q8 cache

28tok/sest.

1627 tok/schat, coding
Devstral Small 2507

Mistral AI

Q5_K_Mfits21.5 GB43k79k with q8 cache

28tok/sest.

1283 tok/schat, coding
Magistral Small 2506

Mistral AI

Q5_K_Mfits21.5 GB40k

28tok/sest.

1283 tok/schat, coding
Magistral Small 2507

Mistral AI

Q5_K_Mfits21.5 GB40k

28tok/sest.

1283 tok/schat, coding
Magistral Small 2509

Mistral AI

Q5_K_Mfits21.5 GB43k79k with q8 cache

28tok/sest.

1264 tok/schat, coding
Qwen 3.5 122B-A10B

Alibaba Qwen

UD-Q2_K_XLoffload40.7 GB–

16tok/sest.

2075 tok/schat
Qwen3 Coder Next

Alibaba Qwen

Q4_K_Moffload46.9 GB–

18tok/sest.

4694 tok/schat
Mistral Small 4 119B

Mistral AI

UD-Q2_K_XLoffload39.0 GB–

22tok/sest.

2354 tok/schat, coding
GLM 4.5 Air

Z.ai

Q2_Koffload48.8 GB–

4tok/sest.

712 tok/s–
Qwen3 Next 80B-A3B

Alibaba Qwen

Q4_K_Moffload46.9 GB–

22tok/sest.

5586 tok/schat, coding
Nemotron 3.5 Lightning 30B-A3B

NVIDIA

UD-Q6_K_XLoffload33.7 GB–

24tok/sest.

7916 tok/schat, coding, long documents
Kimi Linear 48B-A3B

Moonshot AI

Q4_K_Moffload29.2 GB–

55tok/sest.

6467 tok/schat, coding, long documents
Qwen3 32B

Alibaba Qwen

Q4_K_Moffload27.3 GB16k30k with q8 cache

9tok/sest.

678 tok/schat
DeepSeek R1 Distill Qwen 32B

DeepSeek

Q4_K_Moffload27.4 GB16k30k with q8 cache

9tok/sest.

777 tok/schat
granite 4.2 30b

IBM

Q4_K_Moffload25.7 GB23k42k with q8 cache

12tok/sest.

926 tok/schat
DeepSeek V4 Flash

DeepSeek

UD-Q2_K_XLwon't run93.8 GB––––
GLM 5.3 Flash

Z.ai

UD-Q2_K_XLwon't run102.6 GB––––
Qwen 3.8 Flash Next

Alibaba Qwen

UD-Q2_K_XLwon't run75.2 GB––––
MiniMax M2.7

MiniMax

UD-Q2_K_XLwon't run78.8 GB––––
Nemotron 3 Super 120B-A12B

NVIDIA

UD-Q2_K_XLwon't run52.2 GB––––
gpt-oss 120B

OpenAI

Q4_K_Mwon't run60.5 GB––––
Llama 3.3 70B

Meta

Q4_K_Mwon't run50.5 GB––––
DeepSeek R1

DeepSeek

Q4_K_Mwon't run379.7 GB––––
DeepSeek R1 0528

DeepSeek

Q4_K_Mwon't run380.2 GB––––
DeepSeek R1 Distill Llama 70B

DeepSeek

Q4_K_Mwon't run50.5 GB––––
DeepSeek V3 0324

DeepSeek

Q4_K_Mwon't run379.7 GB––––
DeepSeek V3.1

DeepSeek

Q4_K_Mwon't run380.6 GB––––
DeepSeek V3.2

DeepSeek

Q4_K_Mwon't run380.6 GB––––
DeepSeek V4 Flash 0731

DeepSeek

MXFP4won't run149.2 GB––––
DeepSeek V4 Flash Vision Exp

DeepSeek

MXFP4won't run158.0 GB––––
Devstral 2 123B Instruct 2512

Mistral AI

Q8_0won't run135.6 GB––––
GLM 4.5V

Z.ai

Q4_K_Mwon't run65.9 GB––––
GLM 4.6V

Z.ai

Q4_K_Mwon't run72.2 GB––––
GLM 4.7

Z.ai

Q4_K_Mwon't run214.0 GB––––
GLM 5.1

Z.ai

MXFP4won't run423.6 GB––––
GLM 5.2

Z.ai

UD-Q4_K_Mwon't run437.5 GB––––
GLM 5.3

Z.ai

Q8_0won't run750.0 GB––––
Kimi K2

Moonshot AI

Q4_K_Mwon't run581.2 GB––––
Kimi K2 Instruct 0905

Moonshot AI

Q4_K_Mwon't run581.6 GB––––
Kimi K2 Thinking

Moonshot AI

Q4_K_Mwon't run581.6 GB––––
Kimi K2.5

Moonshot AI

Q4_K_Mwon't run581.6 GB––––
MiniMax M2

MiniMax

Q4_K_Mwon't run137.5 GB––––
MiniMax M2.1

MiniMax

Q4_K_Mwon't run137.5 GB––––
MiniMax M2.5

MiniMax

MXFP4won't run123.9 GB––––
MiniMax M3

MiniMax

UD-Q4_K_Mwon't run251.4 GB––––
Mistral Medium 3.5 128B

Mistral AI

Q4_K_Mwon't run81.6 GB––––

Specifications

From the manufacturer's published specifications; sources are linked where we have them.

GPU
GeForce RTX 4090
Memory
24 GB GDDR6X
Memory bus
384-bit
Memory bandwidth
1,008 GB/s
FP16 compute
165 TFLOPS
Software
CUDA
Board power
450 W
PCIe generation
Gen 4
PCIe lanes
x16

Compare with

Questions

How much memory does the NVIDIA GeForce RTX 4090 have for AI models?
The NVIDIA GeForce RTX 4090 has 24 GB of GDDR6X memory with 1,008 GB/s of bandwidth, which sets how fast it generates text.
What is the best model the NVIDIA GeForce RTX 4090 can run locally?
Qwen 3.8 27B (Q5_K_M) fits fully in its memory at 32k context and generates at about 27 tok/s, an estimate from its memory bandwidth. 48 of the 113 models we track fit at that context.
Can the NVIDIA GeForce RTX 4090 run DeepSeek V4 Flash?
No: even at UD-Q2_K_XL, DeepSeek V4 Flash needs about 94 GB at 32k context, more than the NVIDIA GeForce RTX 4090 and system RAM can hold.
How much does the NVIDIA GeForce RTX 4090 cost in Germany?
The best new price for the NVIDIA GeForce RTX 4090 in Germany is 3.955 € at eBay; used, it starts at 2.800 € (asking price).

Speeds on this page are for 32k context. See the best hardware for local AI at every budget.