FitMyLLM
NVIDIA· Blackwell 2.0

NVIDIA GeForce RTX 5090 D V2

Running LLMs on the GeForce RTX 5090 D V2 — the long read: which models fit at which quantisation, and the settings worth changing. · Or what else $1,999 buys

VRAM
24 GB
High-end
Bandwidth
1340
GB/s
Models Q4
312/498
63%
7B Q4 speed
~153
Blazing
Model coverage at Q463% of all
Estimated speed· By model size at Q4

Average speeds at Q4 quantization. Actual performance varies by model architecture and context length.

3B
~357
tok/s
7B
~153
tok/s
14B
~77
tok/s
32B
~34
tok/s
70B
—
39.4GB needed
Measured rig reports

We rent the machine and time every model on it: decode, VRAM peak, concurrency, watts and cost per million tokens. Yours may already be one of them — and two are free to read in full.

See the reports →
Specifications
VRAM
24 GB
Bandwidth
1340 GB/s
FP16 Compute
104.8 TFLOPS
TDP
575W
Memory
GDDR7
Architecture
Blackwell 2.0
CUDA Cores
21,760
Tensor Cores
680
PCIe
Gen 5 x16
MSRP
$1,999
312
Fast models · >30 tok/s
Real-time chat speed
312
Usable · >10 tok/s
Comfortable for all tasks
312
Total compatible
Fit in VRAM at Q4
Where to buy

MSRP $1,999

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

Don’t want to buy?

Test GeForce RTX 5090 D V2 (or anything bigger) without committing. Pay by the second, cancel anytime.

Spin up in ~60s. Pay by the second. Cancel anytime.

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

Compatible models· 312
S
Falcon-H1R Tiny 90M0.09B
Falcon·256K ctx· chat· reasoning
11911
tok/s · 2% VRAM
›
S
SmolLM2 135M0.135B
Smollm·2K ctx· chat
7941
tok/s · 2% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
Embedding·8K ctx· chat
7657
tok/s · 2% VRAM
›
S
GPT-2 124M0.14B
Gpt2·1K ctx· chat
7657
tok/s · 2% VRAM
›
S
SmolVLM 256M0.256B
Smollm·8K ctx· chat· vision
4188
tok/s · 3% VRAM
›
S
Gemma 3 270M0.27B
Gemma·32K ctx· chat
3970
tok/s · 3% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
Embedding·8K ctx· embedding· multilingual
3515
tok/s · 3% VRAM
›
S
bge-large-en-v1.5 335M0.335B
Embedding·1K ctx· chat
3200
tok/s · 3% VRAM
›
S
mxbai-embed-large-v10.335B
Embedding·1K ctx· embedding
3200
tok/s · 3% VRAM
›
S
Snowflake Arctic Embed L0.335B
Embedding·1K ctx· embedding
3200
tok/s · 3% VRAM
›
S
LFM2 350M0.35B
Lfm·125K ctx· chat· tool_use
3063
tok/s · 3% VRAM
›
S
SmolLM2 360M0.36B
Smollm·8K ctx· chat
2978
tok/s · 3% VRAM
›
S
GPT-2 Medium 345M0.38B
Gpt2·1K ctx· chat
2821
tok/s · 3% VRAM
›
S
Qwen 2.5 0.5B0.5B
Qwen·32K ctx· chat
2144
tok/s · 3% VRAM
›
S
SmolVLM 500M0.5B
Smollm·8K ctx· chat· vision
2144
tok/s · 3% VRAM
›
S
Falcon-H1 0.5B0.52B
Falcon·128K ctx· chat
2062
tok/s · 3% VRAM
›
S
BGE-M30.568B
Embedding·8K ctx· embedding
1887
tok/s · 3% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
Embedding·8K ctx· embedding· multilingual
1887
tok/s · 3% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
Embedding·32K ctx· embedding· multilingual
1787
tok/s · 4% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
Falcon·32K ctx· chat· reasoning
1787
tok/s · 4% VRAM
›
S
Falcon Perception 0.6B0.6B
Falcon·4K ctx· vision
1787
tok/s · 4% VRAM
›
S
Qwen 1.5 0.5B0.62B
Qwen·32K ctx· chat
1729
tok/s · 4% VRAM
›
S
LFM2 700M0.74B
Lfm·125K ctx· chat· tool_use
1449
tok/s · 4% VRAM
›
S
Qwen3 0.6B0.75B
Qwen·32K ctx· chat· reasoning
1429
tok/s · 4% VRAM
›
S
GPT-2 Large 774M0.81B
Gpt2·1K ctx· chat
1323
tok/s · 4% VRAM
›
S
Qwen 3.5 0.8B0.87B
Qwen·256K ctx· chat· coding· multilingual
1232
tok/s · 4% VRAM
›
S
InternVL3 1B0.94B
Other·32K ctx· chat· vision
1140
tok/s · 4% VRAM
›
S
MiniCPM5 1B1.08B
Minicpm·128K ctx· chat· reasoning· multilingual
993
tok/s · 5% VRAM
›
S
K2 Horizon 0.9B1.08B
Other·128K ctx· chat· coding· reasoning
993
tok/s · 5% VRAM
›
S
TinyLlama 1.1B1.1B
Llama·2K ctx· chat
975
tok/s · 5% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
Lfm·122K ctx· chat· reasoning· tool_use
893
tok/s · 5% VRAM
›
S
Llama-3.2-1B1.2B
Llama·4K ctx· chat
893
tok/s · 5% VRAM
›
S
LFM2 1.2B1.2B
Lfm·125K ctx· chat· tool_use· multilingual
893
tok/s · 5% VRAM
›
S
Zamba2 1.2B1.2B
Other·4K ctx· chat
893
tok/s · 5% VRAM
›
S
Ling 3.0 Tiny7.89BMoE
Other·128K ctx· chat· coding· reasoning
825
tok/s · 22% VRAM
›
S
EXAONE-4.0-1.2B1.3B
Exaone·64K ctx· chat
825
tok/s · 5% VRAM
›
S
OPT 1.3B1.3B
Opt·2K ctx· chat
825
tok/s · 5% VRAM
›
S
MiniCPM-V 4.61.3B
Other·256K ctx· chat· vision
825
tok/s · 5% VRAM
›
S
DeepSeek Coder 1.3B1.35B
Deepseek·16K ctx· coding
794
tok/s · 5% VRAM
›
S
Phi-1 1.3B1.42B
Phi·2K ctx· coding
755
tok/s · 6% VRAM
›
Next step

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your GeForce RTX 5090 D V2.

What this card is worth

RTX 5090 D V2 holds 312 of the models in our catalogue and is, in practice, a IQ4_XS card — the largest it takes is Nanbeige4.3 33B at IQ4_XS.

Tokens/sec per $100
6.1
8B at Q4_K_M, so cards compare like for like
VRAM per $100
1.20 GB
what memory costs on this card
The biggest it takes
Nanbeige4.3 33B33.12B · IQ4_XS21.5 GB30 tok/sest
DeepSeek Coder 33B33B · IQ4_XS21.6 GB30 tok/sest
DeepSeek-R1-Distill-Qwen-32B32.8B · IQ4_XS21.5 GB30 tok/sest
Qwen3 32B32.8B · IQ4_XS21.5 GB30 tok/sest
Qwen2.5-32B32.5B · IQ4_XS21.3 GB30 tok/sest
Qwen 2.5 Coder 32B32.5B · IQ4_XS21.3 GB30 tok/sest
QwQ-32B32.5B · IQ4_XS21.3 GB30 tok/sest
Granite Switch 4.1 30B Preview32.24B · IQ4_XS21.1 GB31 tok/sest
Against cards of similar memory
M3 Max (36GB)27 GB$2,4991.5 tok/s per $100
M3 Pro (36GB)27 GB$1,9991.2 tok/s per $100
RTX 509032 GB$1,9997.9 tok/s per $100
RTX 5090 D32 GB$1,9997.9 tok/s per $100
RTX 5090 D V224 GB$1,9996.1 tok/s per $100

Neighbours in memory rather than in price: memory decides whether a card can do the job at all, so two cards of the same size at different prices is the comparison you are making. Ordered by memory, then bandwidth — not by the value column, which is worked out from bandwidth and price alone and therefore rewards a cheap card whatever its software stack does to that bandwidth in practice. Read it as one input, not as a ranking.

Device under test

NVIDIA GeForce RTX 5090 D V2 — 24 GB VRAM.

GeForce RTX 5090 D V2 spec
Brand
NVIDIA
VRAM
24 GB GDDR7
Bandwidth
1340 GB/s
FP16 Compute
104.8 TFLOPS
FP32 Compute
104.8 TFLOPS
CUDA Cores
21,760
Tensor Cores
680
TDP
575 W
Architecture
Blackwell 2.0
MSRP
$1999
AI capability
312/ 498 models @ Q4

With 24 GB VRAM and 1340 GB/s bandwidth, this GPU handles models up to 30.5B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~153 tok/s.

Rent it — live prices
Checking what the clouds are charging…
Top models for GeForce RTX 5090 D V2
312 fit · showing 20
ModelSizeVRAM Q4tok/sAvg
Qwen3 30B A3B30.5B19.1 GB35786.2
Qwen3-Coder 30B-A3B30.5B19.1 GB32536.9
Qwen3-30B-A3B Instruct 250730.5B19.1 GB32543.0
MPT-30B30B18.8 GB3626.8
OPT 30B30B18.8 GB366.3
Qwen3-Omni 30B-A3B30B18.8 GB35740.6
Granite 4.1 30B30B18.8 GB3622.6
Muse Glimmer 30B29.78B18.7 GB36233.5
Granite 4.2 30B29.28B18.4 GB3748.9
TranslateGemma 27B28.84B18.1 GB3738.6
Raon-1 mini (preview)28.78B18.1 GB24981.3
PaliGemma 2 28B28B17.6 GB3838.6
ERNIE 4.5 VL 28B A3B Thinking28B17.6 GB357—
Qwen3.5-27B27.8B17.5 GB3959.4
Qwen 3.8 27B27.78B17.5 GB39221.9
gemma-3-27b27.4B17.2 GB39121.7
gemma-2-27b27.2B17.1 GB3934.6
Qwen 3.6 27B27B17.0 GB4054.8
Gemma 4 26B A4B26B16.4 GB268174.3
Aria 25B A3.9B25.3B16.0 GB27564.8