NVIDIA · Consumer · Blackwell (GB203)

NVIDIA GeForce RTX 5080

Only 16 GB; the rumoured 24 GB RTX 5080 SUPER has not launched (NVIDIA's compare page lists no SUPER cards).

VRAM

16 GB

GDDR7

Memory bandwidth

960 GB/s

FP16 compute

112.6 TFLOPS

dense tensor

Price

$999

launch price; no current price

What fits at 4-bit

The 4-bit file plus 8,192 tokens of context and 1.5 GB of overhead, against the VRAM of one card. Speeds are measured medians where someone has published a run at that card count; a blank speed means no run, not a slow one. Machine pages carry estimates for the gaps.

Model	Needs	1 card (16 GB)
Qwen3.5 4B4B · Q4_K_M	4.3 GB	fits
Llama 3.1 8B Instruct8B · Q4_K_M	7.1 GB	45 tok/smedian of 3 runs
Qwen3 8B8.2B · Q4_K_M	7.3 GB	129 tok/smedian of 1 run
Gemma 4 E4B IT8B · Q4_K_M	6.5 GB	fits
Granite 4.2 8B8.8B · Q4_K_M	7.9 GB	fits
Qwen3.5 9B9B · Q4_K_M	7.0 GB	fits
Gemma 4 12B IT11.95B · Q4_K_M	8.2 GB	fits
Ministral 3 14B Reasoning (2512)13.9B · Q4_K_M	10.4 GB	fits
Phi-414.7B · Q4_K_M	11.3 GB	fits
gpt-oss-20b21B, 3.6B active · MXFP4 (native)	12.5 GB	189 tok/smedian of 4 runs

Measured speeds

One card

Single-user runs on 1 card
Model	Quant	Backend	Generation	Prompt	Context	Source
gpt-oss-20b	MXFP4 MoE	llama.cpp (build not stated) CUDA	249 tok/s	11,251 tok/s2,048-token prompt	0	llama.cpp benchmark discussion2026-02
gpt-oss-20b	MXFP4 MoE	llama.cpp b6210 CUDA	205 tok/s	7,477 tok/s2,048-token prompt	0	llama.cpp benchmark discussion2025-08
gpt-oss-20b	MXFP4	llama.cpp b8233	172 tok/s	9,146 tok/s4,096-token prompt	4,096	review2026-03
gpt-oss-20b	MXFP4	Ollama (version not stated)	141 tok/s	5,604 tok/s	0	review2025-10
Llama 3.1 8B Instruct	Q4_K_M	Ollama (version not stated)	135 tok/s	28,928 tok/s	0	review2025-10
Llama 3.1 8B Instruct	Q4_K_M	LocalScore (llamafile; version not stated)	45 tok/s	4,428 tok/s	0	LocalScore2026-10
Llama 3.1 8B Instruct	Q4_K_M	LocalScore (llamafile; version not stated)	43 tok/s	3,963 tok/s	0	LocalScore2026-10
Llama 3.1 8B Instruct	Q8_0	Ollama (version not stated)	91 tok/s	30,170 tok/s	0	review2025-10
Qwen3 8B	Q4_K	llama.cpp (build not stated)	129 tok/s	6,410 tok/s4,096-token prompt	4,096	review2026-03

Speculative decoding, several users, or long context

Kept apart because each measures a different workload from one person chatting near the start of a conversation.

Other runs on this card
Model	Quant	Backend	Generation	Prompt	Context	Measured on	Source
gpt-oss-20b	MXFP4	llama.cpp b8233	161 tok/s	7,561 tok/s16,384-token prompt	16,384	RTX 5080	review2026-03
gpt-oss-20b	MXFP4	llama.cpp b8233	149 tok/s	5,907 tok/s32,768-token prompt	32,768	RTX 5080	review2026-03
Qwen3 8B	Q4_K	llama.cpp (build not stated)	94 tok/s	4,024 tok/s16,384-token prompt	16,384	RTX 5080	review2026-03
Qwen3 8B	Q4_K	llama.cpp (build not stated)	73 tok/s	1,943 tok/s32,768-token prompt	32,768	RTX 5080	review2026-03

Specifications

Memory: 16 GB GDDR7
FP16 basis: Dense FP16/BF16 tensor with FP32 accumulate; FP16 accumulate 225.1 dense (450.2 sparse).
Board power: 360 W
Slots: 2
Length: 304 mm
PCIe: Gen 5 x16
NVLink: No
Cooling: Open-air
Software: CUDA, Vulkan
Released: 2025
Availability: sold new
Price basis: Founders Edition removed from NVIDIA's store in Sept 2026 (NVIDIA: out of stock, not discontinued); launch MSRP $999. Partner cards still sold.
Maker page: nvidia.com

Sources

nvidia.com: 16 GB GDDR7 256-bit, 360 W, FE 304 mm 2-slot, PCIe Gen 5
images.nvidia.com: NVIDIA RTX Blackwell whitepaper appendix tables: memory, bandwidth, TGP, PCIe gen, dense/sparse FP16 tensor TFLOPS
nvidia.com: NVIDIA GeForce compare page: RTX 50 series memory bandwidth row; no RTX 50 SUPER listed
nvidianews.nvidia.com: NVIDIA launch pricing: RTX 5090 $1,999, RTX 5080 $999, RTX 5070 Ti $749
tomshardware.com: RTX 5090/5080 Founders Edition removed from NVIDIA's store in Sept 2026; NVIDIA says still in production
tweaktown.com: RTX 50 SUPER reported delayed indefinitely