How to cut your LLM API bill

The biggest savings come from not sending every request to the most expensive model. After that: limit output length, reuse cached prompts, set budgets so mistakes can't run, use credits you already have, and measure the cost of every request so you know where the money goes.

Start free Read the docs

Request req_8f2c41b9

30 Oct 2026, 14:02:11 SGT · key production-api

Asked forauto

Served byclaude-haiku-5-5 on Anthropic

Input1,840 × $0.10 / 1M

Output212 × $0.50 / 1M

Cost$0.000290

First token · total410 ms · 1.9 s

Six ways, biggest first

Match the model to the job: Small models handle classification, extraction and short summaries; send only the hard work to flagships.
Cap the output: Output tokens often cost 3 to 6 times as much as input. Set max_tokens to what you actually need.
Cache long prompts: Providers that support prompt caching charge much less for repeated context like system prompts and documents.
Set budgets: A per-key monthly budget turns a runaway loop into a stopped request instead of an invoice.
Use credits you have: Cloud startup credits work through your own Azure, Bedrock or Vertex keys.
Measure per request: Cost per request, per feature and per key shows which part of the product to optimise.

An engraving of a wall-mounted telephone transmitter from a 1906 telephone handbook. — From a 1906 handbook on telephone exchanges.
The practical telephone handbook and guide to the telephonic exchange, 1906. Internet Archive Book Images, no known copyright restrictions, via Wikimedia Commons.

What a token costs, by size of model

Prices per million tokens on DigitSpace, 30 Oct 2026.

Model	Input	Output
ministral-3b-2512	$0.10	$0.10
gpt-6-luna	$0.10	$0.50
claude-haiku-5-5	$0.10	$0.50
claude-sonnet-5-5	$2.00	$10.00
gpt-5.5	$5.00	$30.00
claude-fable-5-1	$10.00	$50.00

Current prices for every model are on the models page.

Questions

A telephone exchange room around 1900, with a long wooden switchboard, operator stools and a desk. — General Telephone Company of the Northwest, exchange room, about 1900, public domain, cropped, via Wikimedia Commons.

How much can routing save?

It depends on your traffic. If most requests are simple, moving them from a flagship to a small model cuts their cost by 10 to 100 times; measure your own mix first.

Does a gateway cost extra?

DigitSpace adds 5% on credit top-ups and nothing on tokens; with your own provider keys it's free.

Live in a minute.

Create a key, change the base URL, send a request. No card needed with your own provider keys.

Start free