NVIDIA24 GB VRAM936 GB/s$0.25/hr

SmolLM2 1.7B Instruct on RTX 3090 24GB: Does It Fit, How Fast, What It Costs

SmolLM2 1.7B Instruct at FP16 needs 3.4 GB for its weights and 16.3 GB in total at 8K context with 8 concurrent users. That means it fits on a single RTX 3090 24GB. Every figure is computed by tokcalc's open-source capacity model.

Fits?
Yes, 1 GPU
16.3 GB needed / 24 GB available
Decode speed
178.9
tok/s per stream
Aggregate
2147
tok/s @ 8 concurrent
Cost / 1M tokens
$0.0323
1× RTX 3090 24GB
What changes as you add RTX 3090 24GBs
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× RTX24 GB16.3 GB178.921471.60 s$0.0323✓
2× RTX48 GB16.3 GB304.27301815.7 ms$0.0190✓
4× RTX96 GB16.3 GB608.429203422.8 ms$0.009512✓
8× RTX192 GB16.3 GB1217116813226.4 ms$0.004756✓
Concurrency vs context length
On 1× RTX 3090 24GB at FP16. KV cache scales linearly with context, so this is the number of users your GPUs can actually serve.
ContextConcurrent usersKV per requestFits?
4K25824.6 MB✓ fits
8K121.61 GB✓ fits
33K36.44 GB✓ fits
66K112.88 GB✓ fits
131K025.77 GB✗ does not fit
Other GPUs that run SmolLM2 1.7B Instruct
Same workload, same quantization — ranked by value (tokens per dollar-hour).
GPUGPUsAggregate tok/sCost / 1M tok
RTX 5090 32GB1×4111$0.0338
RTX 4090 24GB1×2312$0.0480
AMD MI300X 192GB1×12159$0.0571
V100 SXM2 32GB1×2065$0.0673
A100 40GB SXM41×3556$0.0703
AMD MI325X 288GB1×13765$0.0706
Frequently asked questions

Can SmolLM2 1.7B Instruct run on RTX 3090 24GB?

Yes. SmolLM2 1.7B Instruct at FP16 needs 16.3 GB including the KV cache, and RTX 3090 24GB has 24 GB — so it fits on a single GPU.

How fast is SmolLM2 1.7B Instruct on RTX 3090 24GB?

178.9 tokens/sec per stream and 2147 tok/s aggregate with continuous batching at 8 concurrent users. Time to first token is 1.60 s, and inter-token latency is 5.6 ms.

What does SmolLM2 1.7B Instruct on RTX 3090 24GB cost per million tokens?

$0.0323 per 1M output tokens at 1× RTX 3090 24GB ($0.2500/hr). That is the marginal hardware cost — add idle capacity, ops, and egress for the real figure.

How much VRAM does SmolLM2 1.7B Instruct need on RTX 3090 24GB?

3.4 GB for the weights at FP16, plus 12.9 GB of KV cache for 8 concurrent requests at 8K context — 16.3 GB total. Every extra concurrent user at long context costs more, so leave headroom.

Is RTX 3090 24GB the right choice for SmolLM2 1.7B Instruct?

It works. Compare $0.0323 per 1M tokens against the alternatives below — the cheapest option that fits the same workload is 2× RTX 3080 10GB at $0.0187, and the fastest is 1× Cerebras CS-3 (WSE-3) at 45882 tok/s.

Adjust the workload yourself

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

Open in calculator