The real cost of LLM inference
Free, source-cited calculators that answer one question with real numbers: is it cheaper to pay for the API or self-host an open-weight model? Break-even volume, monthly TCO, cost per token — with a dated price dataset you can override.
$60API / month*
vs
$348self-hosting / month*
→
69.6Mbreak-even / month
*Example: 10M input + 2M output tokens/mo, Claude Sonnet-class API vs Llama-70B on a rented A100 at 30% utilization. Run your own numbers →
Pick a calculator
Compare
API vs self-hosting, break-even & cost curves
API cost
Token cost, monthly spend, provider comparison
Self-hosting
GPU TCO, $/token, VRAM & model fit
Usage
Token math, context cost, throughput planning
Data
Dated price dataset: APIs, open models, GPUs
The central API vs self-hosting comparator is live; the remaining tools in each pillar are rolling out.
Why another LLM cost calculator?
Most are throwaway JavaScript widgets: not indexable, prices undated and unsourced, no methodology, no named author. LLMTCO is built the opposite way:
- Real numbers, server-rendered. Every tool ships a worked default — no blank boxes, no "click to calculate".
- Dated, sourced prices. Each API and GPU price carries a verification date and a link to the official page. Every price is also an input you can override.
- Transparent math. Every formula is published on the methodology page and verified against known examples.
- Shareable scenarios. Copy a link that reproduces your exact inputs to send to a teammate.