Anyray ← back to site

LLM cost calculator

Estimate what your engineering org spends on LLM inference today — and how much a self-hosted gateway with real-time optimization can take off the bill. Change the inputs; the numbers update live.

01Estimate your savings

Engineers whose assistants, agents, or jobs hit an LLM.

$

Provider spend per dev (tokens across all models).

60%

Anyray typically cuts inference spend around 60%. Drag to model a more conservative case.

Current annual spend
$120,000
$10,000 / month
Estimated annual savings
$72,000
$6,000 / month
Optimized annual spend
$48,000
$4,000 / month
Savings per developer / year
$1,440
$120 / month

Estimate only. Actual savings depend on your workload mix, prompt sizes, and cache hit rate — run a demo to measure against real traffic.

02How the estimate works

The math is deliberately simple so you can sanity-check it:

The savings rate is the one assumption worth scrutinizing. Anyray's real-time optimizer typically removes around 60% of inference cost, but the honest number for your org depends on how much redundant context your agents send and how repetitive your traffic is. The slider lets you model a conservative floor.

03Where the savings come from

Inference cost is just tokens × per-model price. A gateway on the path of every request drives that down four ways, deterministically and with no change to your model:

LeverWhat it doesWhy it saves
Prompt & context compressionTrims redundant context before it reaches the providerCoding agents resend large context every turn — most of it is repeated
Tool pruningDrops tool definitions the request won't useUnused tool schemas are billed as input tokens on every call
Semantic cacheServes repeated or near-identical calls from cacheAgent loops and CI jobs repeat the same calls constantly
RoutingSends each call to the cheapest capable providerThe same request can cost very differently across providers

Compression is deterministic and bounded by quality guardrails, and the gateway fails open — if the optimizer is slow or down, requests pass through untouched. Prompt and response content stays in your environment; it's encrypted at rest and never logged in plaintext.

04Assumptions & method

05See it on real traffic