LLM Cost Calculator

Each request is split into input, cached input, cache writes, output, reasoning and tool fees, each priced at the model's published rate, then multiplied by your monthly volume. Retries add volume, and the batch share gets the provider's published batch discount.

From the live LLM API model catalogue, which powers every model page, refreshed weekly. Batch discounts, max output and promos come from each provider's official pricing page, with the date we last verified them.

Cached input is usually billed at about a tenth of the normal input price. The calculator applies your cache-hit % to the repeated part of each request — system prompt, retrieved context and conversation history — and bills misses at the cache-write price where the provider charges one.

Some providers (OpenAI, Anthropic, Google, Mistral, Alibaba) price asynchronous batch requests 50% lower. Set a batch share for jobs that can wait; models without a published batch price get no discount.

They bill the tokens they think with, usually at the output rate. Add reasoning tokens in the advanced settings and choose an effort level; the cost-per-task column also shows the effect from benchmark runs.

Some models charge a higher rate once a single prompt passes a threshold, such as 200K or 272K tokens. The whole call is then billed at the higher tier. The calculator applies it per call when your scenario crosses the line.

Every turn re-sends the whole conversation so far, so input grows with each turn. Caching the repeated history is the main way to keep it down.

It is exact for OpenAI models, which publish their tokenizer. For other models it is an estimate of about 4 characters per token and is labelled approx.

Because the provider hasn't published that figure — a cache price, a batch discount or output tokens per task. We never estimate a price, so the line stays empty.

Generating tokens is the expensive part of running a model: each output token needs a full forward pass, while input tokens are processed in parallel. Most providers charge three to five times more per output token.

Weekly, from the live LLM API model catalogue — the same data that powers every model page, so prices never drift between pages. Provider fees and discounts carry the date we last verified them against the official pricing page.

Yes. Copy share link puts your whole setup — models, volume, advanced settings — into the URL, and Download CSV exports the current results table.

Yes. No sign-up is needed. When you are ready, create an API key and your first deposit is doubled.