IFX Optimizer · CLI
Find the money you're wasting on AI.
The index tells you what inference costs. The Optimizer tells you what youare overpaying. It scans your codebase, prices every LLM call against today's live IFX rates, and flags the waste — frontier models on trivial tasks, missing caching, silent retries.
npx @ifx/optimizer scan .Proof
We pointed it at 22 of our own apps. One was bleeding 32%.
Most were already lean — and the tool said so, instead of inventing savings. But one ran Claude Sonnet for classification, intent, and extraction: jobs a capable-tier model handles at a fifth of the price. The Optimizer found it, priced it from the live index, and showed the fix.
How it works
Detect. Price. Prove.
- 1
Drop in your repo.
One command scans every OpenAI, Anthropic, and Google call — even when the model is set through a config or a helper function.
- 2
Priced against the live index.
Each call is costed using today's real IFX rates. Over-provisioned frontier models get pointed at the cheapest model that still clears the task's capability bar.
- 3
Proven, not guessed.
Connect your usage export and modeled estimates become measured dollars. Model downgrades ship as a pull request only after an eval confirms quality holds.
Pricing
Free to find it. Pro to prove it.
Free
$0- Local repo scan
- Pricing vs a bundled snapshot
- Runtime duplication & retry rates
- Source never leaves your machine
Pro
$29 / dev / mo- Live daily IFX feed
- Measured usage merge
- Eval-gated downgrade PRs
- Dashboard, history & teams
Source code never leaves your machine. Pro sync uploads findings metadata only — never source or prompts.
See what your app is wasting.
Runs locally. No signup to scan.
npx @ifx/optimizer scan .