Qwen3.8 27B — VRAM requirements & what it costs to run
Bottom line: at q4_k_m with 8k context you need ≈ 18.3 GiB; smallest single accelerator in our catalog: RTX 3090 24GB.
Quantization table (GiB)
| quant | weights | total @8k | need @8k | smallest fit |
|---|---|---|---|---|
| native | 51.75 | 57.97 | 58.0 | A100 80GB |
| bf16 | 51.75 | 57.97 | 58.0 | A100 80GB |
| fp8 | 25.87 | 29.51 | 29.5 | RTX 5090 32GB |
| q8_0 | 27.49 | 31.29 | 31.3 | RTX 5090 32GB |
| q6_k | 21.22 | 24.39 | 24.4 | RTX 5090 32GB |
| q5_k_m | 18.34 | 21.22 | 21.2 | RTX 3090 24GB |
| q4_k_m | 15.65 | 18.27 | 18.3 | RTX 3090 24GB |
| q3_k_m | 12.65 | 14.96 | 15.0 | RTX 4060 Ti 16GB |
| q2_k | 10.83 | 12.97 | 13.0 | RTX 4060 Ti 16GB |
Dataset: modelvram.com open dataset (CC BY 4.0 / MIT), retrieved 2026-10-20. "total @8k" already includes KV cache + runtime overhead per dataset methodology; we extrapolate KV linearly by context (formula).
KV-cache scaling at q4_k_m (total GiB)
| context | total |
|---|---|
| 4k | 18.0 |
| 8k | 18.3 |
| 16k | 18.8 |
| 32k | 19.8 |
| 128k | 25.8 |
Cost: rent vs buy vs API
Cheapest listed rental that fits: RunPod RTX 5090 32GB @ $0.99/h — source: modelvram.com public table (self-stated check 2026-09-29, retrieved 2026-10-20). Partial re-verify 2026-10-20: Lambda A100 40GB $1.99/h CONFIRMED on lambda.ai/pricing (fetched directly); Lambda H100 rows ambiguous on-page ($3.99 vs $4.09 configs) → not adopted; RunPod & Vast pricing pages are JS-rendered, no extractable static figures — verify rates visually before spending
Cross-check on record for this family: vLLM official recipes, retrieved 2026-10-20: FP8 ≈ 200 GB, NVFP4 ≈ 170 GB; Thunder Compute blog, 2026-08-12: native checkpoint ≈ 175 GB, 3-bit GGUF ≈ 131 GB. Datasets differ by quantisation bookkeeping — see method.
Running Qwen3.8 27B with Ollama
Typical q4_k_m GGUF download ≈ 15.65 GiB (weights, dataset figure); expect ≥ 18.3 GiB resident at 8k context. Find tags: ollama.com/search (we do not mirror tag lists — sizes on Ollama are their numbers, ours are dataset-sourced).
FAQ
How much VRAM does Qwen3.8 27B need?
At q4_k_m with an 8k context it needs about 18.3 GiB including KV cache and runtime overhead; the smallest card in our catalog that fits is RTX 3090 24GB. Numbers come from the modelvram.com open dataset (CC BY 4.0, retrieved 2026-10-20) and our published formula (https://llmbudgetlab.com/method/).
Can I run Qwen3.8 27B on a Mac?
Use the calculator: macOS caps GPU-usable memory well below your spec-sheet unified RAM. We count 2/3 of unified (3/4 above 36GB) as a deliberately conservative rule of ours, not an Apple constant — compare required GiB against usable memory, then confirm on your own machine with sysctl iogpu.wired_limit_mb (why).
Is renting a GPU cheaper than buying one for Qwen3.8 27B?
Rental from $0.99/h (RunPod) — plug both into the calculator's break-even field; buy price and electricity are yours to enter because we refuse to guess market prices.
**Qwen3.8 27B** needs ~18.3 GiB VRAM at q4_k_m (8k context) → minimum RTX 3090 24GB. Check & cost-compare: [LLM Budget Lab](https://llmbudgetlab.com/models/qwen3-8-27b/)or grab the SVG badge →
More: builds under $1000 · under $1500 · used GPUs & break-even · RTX 4090 · RTX 5090