How much VRAM Qwen3.6 27B needs, and what it costs to run
Trying to decide what to buy? The best GPU for Qwen3.6 27B →
Cheapest way to rent it right now
$0.01/hr
NVIDIA DGX Spark (GB10 Grace Blackwell) · F16 · spot · ≈ $3/mo at 8h/day
Live price captured Sat, 07 Nov 2026 17:20:07 GMT.Referral link — we may earn a commission at no cost to you; it never changes which card is cheapest.
Just want to use it? Per-token API
$0.60in·$3.60out
per 1M tokens · Qwen3.6 27B on Novita AI · read Oct 29, 2026
No GPU to manage: you pay per token, which suits occasional use. Your prompts run on Novita's servers, not your own machine.
Price as listed on Novita's pricing page on Oct 29, 2026; it may have changed. We may earn a commission at no cost to you.
Rent a GPU or pay per token?
A per-token API charges for every token, so it is cheapest when you use Qwen3.6 27B a little. A rented GPU charges by the hour whether it is busy or idle, so it only pays off once you push enough tokens through it. Set your daily volume and see where the line falls for this model, using the cheapest live rental that fits and that we have a memory-bandwidth figure for (DGX Spark (GB10 Grace Blackwell), F16, $0.013/hr spot) and Novita AI's published rate.
- Pay per token · Novita AI$40.50/mo
$1.35/day · $1.35 per 1M tokens blended at 3:1
- Rent DGX Spark (GB10 Grace Blackwell) · Vast.ai$9.77/mo
$0.33/day · 24.5 GPU-hours needed (2 cards running all day)
At 1M tokens/day renting is cheaper. Below ~9.9K tokens/day the API is cheaper; above it, renting the DGX Spark (GB10 Grace Blackwell) wins.
At a glance · 3:1 input:output
| Tokens/day | API /mo | Rent /mo |
|---|---|---|
| 100K | $4.05 | $0.98 |
| 1M | $40.50 | $9.77 |
| 10M | $405 | $97.75 |
| 100M | $4,050 | $977 |
Cheaper option in green. 30-day month. Break-even: ~9.9K tokens/day.
How we work it out. API cost = input tokens × $0.60 + output tokens × $3.60 per 1M. Rent cost = the hourly price × the GPU time needed, with at least 1 hour billed per day (spin-up, model load and idle time cost money too). GPU time assumes the DGX Spark (GB10 Grace Blackwell) writes ~4 output tokens/sec for Qwen3.6 27B at F16 (our estimate: its 273 GB/s memory bandwidth from the spec sheet, not yet confirmed by a second source ÷ the bytes of weights read per token, × 75% efficiency, one request at a time) and reads prompt tokens 10× faster than it writes. Above 24 GPU-hours a day you would need more than one card.
These are estimates: the tokens/sec figure is our model, not a benchmark, and batching many requests (vLLM, SGLang) can raise a rented card's throughput a lot, which moves the break-even lower. Rental price captured Sat, 07 Nov 2026 23:20:11 GMT; it is a spot price, and spot machines can be interrupted. API price as listed by Novita AI on Oct 29, 2026; API prices change. Referral links: we may earn a commission at no cost to you, and it never changes which option the numbers favour.
Memory needed, by quantisation
At 16k context. Weights + KV cache + ~0.8 GB overhead. Memory is in binary GB (1,024³ bytes), the unit GPU memory is sold in; Hugging Face and Ollama list downloads in decimal GB, about 7% more.
| Quant | Weights | KV cache | Total | Quality |
|---|---|---|---|---|
| F16 | 51.7 GB | 1.0 GB | 53.5 GB | Full precision. Reference quality, twice the size of Q8 for no practical gain in most chat use. |
| Q8_0 | 27.5 GB | 1.0 GB | 29.3 GB | Effectively lossless. Use when VRAM is not the constraint. |
| Q6_K | 21.2 GB | 1.0 GB | 23.0 GB | Very close to Q8 at meaningfully less memory. A safe high-quality pick. |
| Q5_K_M | 18.4 GB | 1.0 GB | 20.2 GB | Small, hard-to-notice quality loss. Good balance. |
| Q5_0 | 17.9 GB | 1.0 GB | 19.7 GB | Older-style 5-bit. Q5_K_M is usually the better pick at the same size. |
| Q4_K_M | 15.7 GB | 1.0 GB | 17.5 GB | The community default. Best quality-per-gigabyte for most people. |
| Q4_0 | 14.7 GB | 1.0 GB | 16.5 GB | Older-style 4-bit, measurably worse than Q4_K_M at a similar size. Avoid unless required. |
| IQ4_XS | 13.7 GB | 1.0 GB | 15.5 GB | Newer 4-bit, smaller than Q4_K_M with comparable quality. Needs a recent llama.cpp. |
| Q3_K_M | 12.6 GB | 1.0 GB | 14.4 GB | Noticeable degradation. Use to fit a larger model that would otherwise not run. |
| Q2_K | 10.8 GB | 1.0 GB | 12.6 GB | Heavy degradation. Almost always better to run a smaller model at Q4_K_M instead. |
| IQ3_XS | 10.7 GB | 1.0 GB | 12.5 GB | Aggressive. Usually better than a smaller model at Q4, but test before trusting it. |
Cards that can run it
Best quantisation each card fits at 16k context, with live rental price where we track one.
How these numbers are produced
- Model shape comes from the model's own config.json — 16 layers, 24 attention heads, 4 KV heads.
- 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.
- Some memory-bandwidth figures are not yet independently verified.Where that's the case we show no tokens/sec at all rather than a number we can't stand behind.
- Rental prices are pulled hourly from provider APIs (last updated Sat, 07 Nov 2026 17:20:07 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
Qwen3.6 27B locally: the all-dense 24GB Qwen
Qwen3.6's dense 27B is the no-fuss 24GB-card option — every parameter fires, nothing to offload, a real 256K context. The VRAM math, the GPU, and how it differs from the 35B-A3B MoE.
How much VRAM do you need?
About 0.57 GB per billion parameters at 4-bit, plus room for context: the rule behind the numbers on this page, with worked examples.
The best GPU for local AI right now
The card with enough VRAM for your model at the lowest price, budget by budget.
Rent a cloud GPU or buy one?
Rent, unless you'll pass about 2,000 hours of use: the break-even and the costs people forget on both sides.