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

GPT-OSS 120B (MoE) on RTX 5090 32GB: Does It Fit, How Fast, What It Costs

GPT-OSS 120B (MoE) at AWQ 4-bit needs 64.3 GB for its weights and 66.7 GB in total at 8K context with 8 concurrent users. That means it needs 4× RTX 5090 32GB. Every figure is computed by tokcalc's open-source capacity model.

Fits?
Yes, 4×
66.7 GB needed / 128 GB available
Decode speed
1242
tok/s per stream
Aggregate
59638
tok/s @ 8 concurrent
Cost / 1M tokens
$0.009316
4× RTX 5090 32GB

RTX 5090 32GB has no NVLink, so 4-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 GB66.7 GB0.0000.000——✗
2× RTX64 GB66.7 GB0.0000.000——✗
4× RTX128 GB66.7 GB124259638428.7 ms$0.009316✓
8× RTX256 GB66.7 GB2485238551229.4 ms$0.004658✓
Concurrency vs context length
On 4× RTX 5090 32GB at AWQ 4-bit. KV cache scales linearly with context, so this is the number of users your GPUs can actually serve.
ContextConcurrent usersKV per requestFits?
4K409309.2 MB✓ fits
8K207618.5 MB✓ fits
33K522.42 GB✓ fits
66K264.83 GB✓ fits
131K139.66 GB✓ fits
Other GPUs that run GPT-OSS 120B (MoE)
Same workload, same quantization — ranked by value (tokens per dollar-hour).
GPUGPUsAggregate tok/sCost / 1M tok
AMD MI300X 192GB1×12969$0.0535
L40S 48GB2×7188$0.0541
AMD MI325X 288GB1×14682$0.0662
A100 80GB SXM41×4894$0.0681
Intel Gaudi 3 128GB1×8075$0.0688
H100 SXM5 80GB1×8198$0.0847
Frequently asked questions

Can GPT-OSS 120B (MoE) run on RTX 5090 32GB?

Yes, but it needs 4× RTX 5090 32GB. At AWQ 4-bit the total footprint is 66.7 GB (64.3 GB of weights plus 2.5 GB of KV cache at 8K × 8 concurrent users), against 128 GB of available VRAM across 4 GPUs.

How fast is GPT-OSS 120B (MoE) on RTX 5090 32GB?

1242 tokens/sec per stream and 59638 tok/s aggregate with continuous batching at 8 concurrent users. Time to first token is 428.7 ms, and inter-token latency is 804.9 µs.

What does GPT-OSS 120B (MoE) on RTX 5090 32GB cost per million tokens?

$0.009316 per 1M output tokens at 4× 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 GPT-OSS 120B (MoE) need on RTX 5090 32GB?

64.3 GB for the weights at AWQ 4-bit, plus 2.5 GB of KV cache for 8 concurrent requests at 8K context — 66.7 GB total. Every extra concurrent user at long context costs more, so leave headroom.

Is RTX 5090 32GB the right choice for GPT-OSS 120B (MoE)?

It works. Compare $0.009316 per 1M tokens against the alternatives below — the cheapest option that fits the same workload is 8× RTX 3080 10GB at $0.004393, and the fastest is 2× Cerebras CS-3 (WSE-3) at 166400 tok/s.

Adjust the workload yourself

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

Open in calculator