Run Llama 3.1 8B on a VPS.
September 2026 · 207 of 280 plans fit · re-ranked daily
The reference 8B: broad ecosystem, solid general chat and RAG, fits an 8 GB plan at Q4. On a CPU-only VPS it needs about 7.4 GB of RAM: 4.9 GB of Q4_K_M weights, 1 GB of KV cache for an 8,192-token context and 1.5 GB for the OS and runtime. 207 of the 280 plans in our index fit; the cheapest comfortable pick is Contabo's Cloud VPS 6 · 6 vCPU · 12 GB at $8.72/mo, streaming an estimated 3–5.9 tok/s.
RAM needed · CPU inference
7.4 GB
- weights · Q4_K_M
- 4.9 GB
- KV cache · 8K context
- 1 GB
- OS + runtime headroom
- 1.5 GB
Other quants: Q8_0 weights 8.5 GB (near-lossless, about half of F16); F16 16.1 GB. Comfortable = 20% headroom over the total.
Model card
- Size
- 8.03B parameters
- Context
- 128K tokens
- Kind
- chat
- Released
- 2024-07
- License
- Llama 3.1 Community License
Facts fetched from Hugging Face on 3 Sept 2026: exact GGUF file sizes, KV geometry from the GGUF header · 5,792,238 downloads.
Best VPS plans for Llama 3.1 8B
ranked by estimated tokens/s per dollar, comfortable fits first
- 1runs comfortably~3–5.9 tok/s$8.72/mo
- 2runs comfortably~5.9–12 tok/s$20.93/mo
- 3runs comfortably~3.9–7.9 tok/s$16.27/mo
- 4runs comfortably~3.9–7.9 tok/s$18.58/mo
- 5runs comfortably~3.9–7.9 tok/s$18.80/mo
- 6runs comfortably~3–5.9 tok/s$14.50/mo
Speed = effective memory bandwidth ÷ active weight bytes (4 GB/s per shared vCPU, 6 per dedicated), shown as a band. Real numbers depend on the host CPU generation, AVX-512/AMX support and how noisy the neighbours are — treat these as order-of-magnitude.
Or buy hardware · 14 reference machines fit
all machines →- Raspberry Pi 5 (16 GB)16 GB · 17 GB/s · ~0.9–1.4 tok/s$305= 35 mo of VPS
- GeForce RTX 5060 Ti 16 GB (card only)16 GB · 448 GB/s · ~48–74 tok/s$805*= 92 mo of VPS
- Mac mini (M6, 16 GB)16 GB · 153 GB/s · ~17–25 tok/s$899= 103 mo of VPS
- GeForce RTX 5070 Ti 16 GB (card only)16 GB · 896 GB/s · ~97–148 tok/s$1,100*= 10+ yrs of VPS
- Mac mini (M6, 32 GB)32 GB · 153 GB/s · ~17–25 tok/s$1,299= 10+ yrs of VPS
- GeForce RTX 5080 16 GB (card only)16 GB · 960 GB/s · ~100+ tok/s$1,500*= 10+ yrs of VPS
Buy · per month
$8.78
$8.47 hardware + $0.31 power
Rent · per month
$8.72
Contabo Cloud VPS 6 · 6 vCPU · 12 GB
Break-even
The Raspberry Pi 5 (16 GB) pays for itself after 36 months of replacing the VPS — and it streams an estimated 0.9–1.4 tok/s against the VPS's CPU-only pace.
* approximate: August 2026 US retail median rather than list price. Local speed = peak bandwidth × 0.7 efficiency (0.35 on CPU-only boards) ÷ active weight bytes; unified-memory machines are assumed to give models 75% of their RAM. GPU cards need a host PC that is not included in the price.
Frequently asked
- How much RAM does Llama 3.1 8B need?
- About 7.4 GB for CPU inference at Q4_K_M: 4.9 GB of weights, 1 GB of KV cache at 8,192 tokens of context, and 1.5 GB of headroom. Longer contexts need more KV cache (128 KB per token for this model).
- What is the cheapest VPS that can run Llama 3.1 8B?
- Contabo Cloud VPS 6 · 6 vCPU · 12 GB (12 GB RAM, 6 vCPU) at $8.72/mo excl. VAT runs it comfortably as of 5 Sept 2026. Contabo Cloud VPS 4 · 4 vCPU · 8 GB at $6.39/mo is a tight fit.
- How fast will Llama 3.1 8B run on a VPS without a GPU?
- Roughly 3–5.9 tok/s on the top pick. Token generation is bound by memory bandwidth — each token streams the full weights once — so more and dedicated vCPUs help; a GPU or Apple-silicon machine is 10–50× faster.