Code Llama 70B.
Meta's largest code-specialised Llama.
1× Nvidia L40S · $0.28/hr.
Most-aggressive quantisation we have a working recommendation for. Lower precision = less VRAM = cheaper hardware, at a small accuracy cost.
Cheapest hosted endpoints.
| Provider | Access | $/M in | $/M out | |
|---|---|---|---|---|
| Ollama | self hosted | — | — | Launch ↗ |
Smaller models distilled from Code Llama 70B.
Lightweight student models trained to mimic Code Llama 70B's outputs.
Variants in the Code Llama family.
Frequently asked.
How do I run Code Llama 70B?
Where can I access Code Llama 70B?
How much does it cost to run Code Llama 70B?
Is Code Llama 70B open-source or proprietary?
Cheapest hardware per quantisation.
Each row is one quantisation tier (the same weights compressed differently). Lower precision → lower VRAM → cheaper hardware, at the cost of small accuracy loss. $/hr refreshed hourly from each provider's API.
| Quantisation | Cheapest GPU config | Total VRAM | Live $/hr | tokens/sec | |
|---|---|---|---|---|---|
|
FP16
FP16 — half precision (default)
|
192 GB | — | — | Compare → | |
|
FP8
FP8 — 8-bit float (Hopper / Blackwell)
|
94 GB | — | — | Compare → | |
|
INT4
INT4 — 4-bit integer (~4× VRAM saving)
|
48 GB | $0.28/hr | — | Compare → |
What it costs per month across providers.
Estimate your monthly bill for Code Llama 70B across every host that publishes per-token pricing. Slide your token volumes; the chart + table re-rank cheapest-first.
No priced API access rows on file for Code Llama 70B yet.
Rent the GPU instead of paying per token.
For an open-weights model like Code Llama 70B, you can rent a GPU and serve inference yourself. The math: cheapest GPU rental × 730 hours/month + your electricity rate × power draw.
Assumes the GPU runs 24/7 at ~85% utilisation. If your traffic is bursty, you'll pay less for the API and probably more for the GPU (idle hours still cost rental). The breakeven analysis lives on the Self-host vs API breakeven tool.