NVIDIA32 GB VRAM1,792 GB/s$0.5/hr

Llama 2 7B on RTX 5090 32GB: Does It Fit, How Fast, What It Costs

Llama 2 7B at FP16 needs 13.5 GB for its weights and 47.8 GB in total at 8K context with 8 concurrent users. That means it needs 2× RTX 5090 32GB. Every figure is computed by tokcalc's open-source capacity model.

Fits?
Yes, 2×
47.8 GB needed / 64 GB available
Decode speed
146.9
tok/s per stream
Aggregate
3526
tok/s @ 8 concurrent
Cost / 1M tokens
$0.0788
2× RTX 5090 32GB

RTX 5090 32GB has no NVLink, so 2-way tensor parallelism runs over PCIe. tokcalc applies a flat 0.85× multi-GPU efficiency factor for this, which is realistic for NVLink-equipped datacenter cards but optimistic for PCIe scale-out — treat the throughput and cost figures for this configuration as an upper bound, and measure before committing.

What changes as you add RTX 5090 32GBs
Tensor-parallel degree vs VRAM, throughput, and cost. Cost scales linearly with GPU count; decode throughput scales near-linearly too, so cost per token stays roughly flat.
GPUsTotal VRAMVRAM neededDecode tok/sAggregate tok/sTTFTCost / 1M tokFits?
1× RTX32 GB47.8 GB0.0000.000——✗
2× RTX64 GB47.8 GB146.935261.09 s$0.0788✓
4× RTX128 GB47.8 GB293.814102559.3 ms$0.0394✓
8× RTX256 GB47.8 GB587.656408294.6 ms$0.0197✓
Concurrency vs context length
On 2× RTX 5090 32GB at FP16. KV cache scales linearly with context, so this is the number of users your GPUs can actually serve.
ContextConcurrent usersKV per requestFits?
4K232.15 GB✓ fits
8K114.29 GB✓ fits
33K217.18 GB✓ fits
66K134.36 GB✓ fits
131K068.72 GB✗ does not fit
Other GPUs that run Llama 2 7B
Same workload, same quantization — ranked by value (tokens per dollar-hour).
GPUGPUsAggregate tok/sCost / 1M tok
AMD MI300X 192GB1×3067$0.2264
AMD MI325X 288GB1×3472$0.2800
A100 80GB SXM41×1157$0.2880
Intel Gaudi 3 128GB1×1909$0.2909
H100 SXM5 80GB1×1938$0.3583
B200 SXM 192GB1×4629$0.3600
Frequently asked questions

Can Llama 2 7B run on RTX 5090 32GB?

Yes, but it needs 2× RTX 5090 32GB. At FP16 the total footprint is 47.8 GB (13.5 GB of weights plus 34.4 GB of KV cache at 8K × 8 concurrent users), against 64 GB of available VRAM across 2 GPUs.

How fast is Llama 2 7B on RTX 5090 32GB?

146.9 tokens/sec per stream and 3526 tok/s aggregate with continuous batching at 8 concurrent users. Time to first token is 1.09 s, and inter-token latency is 6.8 ms.

What does Llama 2 7B on RTX 5090 32GB cost per million tokens?

$0.0788 per 1M output tokens at 2× RTX 5090 32GB ($0.5000/hr). That is the marginal hardware cost — add idle capacity, ops, and egress for the real figure.

How much VRAM does Llama 2 7B need on RTX 5090 32GB?

13.5 GB for the weights at FP16, plus 34.4 GB of KV cache for 8 concurrent requests at 8K context — 47.8 GB total. Every extra concurrent user at long context costs more, so leave headroom.

Is RTX 5090 32GB the right choice for Llama 2 7B?

It works. Compare $0.0788 per 1M tokens against the alternatives below — the cheapest option that fits the same workload is 8× RTX 3080 10GB at $0.0186, and the fastest is 2× Cerebras CS-3 (WSE-3) at 39347 tok/s.

Adjust the workload yourself

Same formulas, your context length, concurrency, and quantisation.

Open in calculator