Which LLM API is cheapest for your workload?

There is no single cheapest LLM API for every workload. Compare the same model, input and output rates, cache conditions, and minimum charges. A low input rate can still cost more when a coding agent produces long answers.

Catalog: Verified on 2026-11-06. OpenRouter listed-price snapshot checked 2026-11-02T18:09:39.794Z. Rates are regenerated from public data, not edited by hand.

The competitor snapshot is awaiting refresh. Comparative counts and exceptions are withheld until it is refreshed; the live Meldport table remains available.

Input-to-output ratio changes the result

For one million input tokens and r million output tokens, cost is input rate + r × output rate. Compare both providers with the same r. If one provider has cheaper input but more expensive output, solve input A + r × output A = input B + r × output B to find the crossover. See worked ratio calculations.

When Meldport is not the cheapest

Check refreshed listed prices before selecting a provider.

Some models are cheaper here only on output, others only on input. Missing matches are excluded, not counted as wins. Matching uses exact legacy model IDs against the suffix of OpenRouter IDs; ambiguous matches, aliases containing ~, and :batch variants are excluded. This does not certify equivalent upstream routing or performance. Inspect the OpenRouter source.

Meldport's shared monthly free allowance has separate eligibility and optional wallet fallback. The paid-rate comparison above does not treat free tokens as an unlimited discount. Media APIs are outside this token-price comparison. Read the migration tradeoffs and check agent compatibility before buying credit.