What it costs to run Gemma 3 27B
Trying to decide what to buy? The best GPU for Gemma 3 27B →
We don't yet have the layer/attention shape for this model — its repository is licence-gated, so the KV-cache figures below can't be computed. Weight sizes are still accurate.
How these numbers are produced
- Model shape comes from the model's own config.json.
- Weight size is parameters × effective bits-per-weight. Those constants are checked against real published quantised file sizes — median error 0.7% across 40 measurements.
- KV cache is 2 × layers × KV-heads × head-dim × context × 2 bytes. Using KV-heads rather than attention heads is what makes this correct for grouped-query attention; treating a GQA model as multi-head overstates the cache by up to 8×.
- Tokens/sec is an estimate, not a benchmark. Generation is memory-bandwidth-bound, so this is bandwidth ÷ weight-bytes derated to 75%. Real throughput depends on your runtime, batch size and kernels. We don't run our own hardware tests — see our editorial policy.
- Rental prices are pulled hourly from provider APIs (last updated Wed, 22 Jul 2026 22:20:08 GMT). Spot/interruptible pricing can change or vanish without notice.
Try it interactively
Drag the context slider and watch the KV cache fill the card
Worth reading before you buy
RTX 5090 vs two RTX 3090s
Two 3090s give you 48 GB but not double the speed for chat — llama.cpp's default multi-GPU mode makes the cards take turns. What the benchmarks actually show.
Cheapest way to serve Llama 70B
"Cheapest" flips on duty cycle and concurrency — and in Europe the electricity bill alone can approach the cost of just renting. Plus the config defaults that silently break a 16k deployment.
What --n-cpu-moe actually does
The trick that makes MoE models 5× faster — except it usually makes them slower, and the famous speedup only happens when the model didn't fit in the first place.