Home › Models › DeepSeek V4 Flash 0731
DeepSeek V4 Flash 0731 — VRAM requirements & what it costs to run
Bottom line: at q4_k_m with 8k context you need ≈ 189.8 GiB; smallest single accelerator in our catalog: Mac 256GB unified.
Quantization table (GiB)
| quant | weights | total @8k | need @8k | smallest fit |
|---|---|---|---|---|
| native | 155.43 | 172.21 | 172.2 | Mac 256GB unified |
| bf16 | 566.58 | 624.48 | 624.5 | needs multi-GPU / offload |
| fp8 | 283.29 | 312.86 | 312.9 | needs multi-GPU / offload |
| q8_0 | 301.0 | 332.33 | 332.3 | needs multi-GPU / offload |
| q6_k | 232.3 | 256.77 | 256.8 | needs multi-GPU / offload |
| q5_k_m | 200.78 | 222.1 | 222.1 | needs multi-GPU / offload |
| q4_k_m | 171.39 | 189.77 | 189.8 | Mac 256GB unified |
| q3_k_m | 138.46 | 153.54 | 153.5 | Mac 256GB unified |
| q2_k | 118.63 | 131.73 | 131.7 | H200 141GB |
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 | 189.4 |
| 8k | 189.8 |
| 16k | 190.4 |
| 32k | 191.8 |
| 128k | 199.8 |
Cost: rent vs buy vs API
Needs ≥189.8 GiB — no single listed rental instance fits (multi-GPU / offload only). 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 DeepSeek V4 Flash 0731 with Ollama
Typical q4_k_m GGUF download ≈ 171.39 GiB (weights, dataset figure); expect ≥ 189.8 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 DeepSeek V4 Flash 0731 need?
At q4_k_m with an 8k context it needs about 189.8 GiB including KV cache and runtime overhead; the smallest card in our catalog that fits is Mac 256GB unified. 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 DeepSeek V4 Flash 0731 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 DeepSeek V4 Flash 0731?
Rental requires multi-GPU — plug both into the calculator's break-even field; buy price and electricity are yours to enter because we refuse to guess market prices.
**DeepSeek V4 Flash 0731** needs ~189.8 GiB VRAM at q4_k_m (8k context) → minimum Mac 256GB unified. Check & cost-compare: [LLM Budget Lab](https://llmbudgetlab.com/models/deepseek-v4-flash-0731/)or grab the SVG badge →
More: builds under $1000 · under $1500 · used GPUs & break-even · RTX 4090 · RTX 5090