An SDK that installs inside your code, lets through only what has to be paid for, and returns a SavingsReport per call: measured and estimated kept apart, counted with your provider's own meter. We publish only what we have measured.
Any endpoint compatible with the OpenAI API, plus the native Anthropic adapter. With a published figure, today, only OpenAI gpt-4o-mini.
The headline is the floor of the 95% interval, never the point.
avoided_calls22%eliminated_tokens43.2%compressed_tokens20.1%Each ledger is published with its own interval; those that do not clear zero are not advertised.run Sep 12, 2026 · d1c8233
Fewer bill headaches. Without changing provider.
Three facts about the package. None of them is a promise.
Zero runtime dependencies
A single ESM bundle. Nothing gets installed next to your code that we did not write; the exact tokenizer is an optional peer.
One SavingsReport per call
Four ledgers that are never summed, the cost with its basis (measured or estimated) and traces per stage. It is the same thing your dashboard reads.
Your traffic still goes direct
The SDK wraps the call inside your process. Your prompts never pass through BiVelio: we only ever receive counts and cost.
What does Savings actually do?
Four ideas, four eight-second animations. That's the whole product.
AI charges you per word
Every request to an AI is paid for by the text going in and out. Savings stands in the middle of that path and lets through only what's needed: whatever you didn't have to pay for drops into your jar.
You don't pay twice for the same thing
An AI re-reads everything you already sent it, turn after turn — and the provider charges for it again. Savings prepares every request so the provider applies its discounted price to the repeats. The discount is the provider's; our part is asking for it properly.
A receipt you can verify
We don't ask for faith. Every saving is measured with the provider's official meter — the same one they bill you with — and lands on your receipt. Anything that can't be checked against the bill is never announced.
We only earn when you save
Our commission comes out of the measured saving, never out of your pocket: most of what you save always stays with you. A month with no savings? Zero commission.
The animations illustrate the idea. The real numbers — measured against live providers — are right below.
How do we know it saves? We measure it — and we publish every trial, one by one.
Every scenario runs twice against the real provider: as-is, and with Savings in front. Both results are counted with the provider's own meter — the same one that prices your bill. Here is what came out, error margin included:
gpt-4o-miniEach step falls into ONE different ledger, so these figures do not add up to each other. A step whose ledger does not clear zero does not appear here: output restriction was measured and does not save, so it is not advertised. Every figure in this run is from gpt-4o-mini: the same lever can cost money on another model, and that is published too.
The test is a fair experiment: the same task twice — without Savings and with Savings — with the provider's own meter counting both.
Translation, no jargon
In this test, the bill went down by at least 23.2%, with near certainty. Most likely it went down by 24.6%.
When there's a range, we always announce the small number — the one statistics guarantees — never the flattering one.
Four ledgers, never summed — each is a separately measured lever, so a dollar can't be double-counted. Reconciliation residual ≈ 0 proves it. These are reference levers: your live rail measures a different four — provider-cached replaces output-restriction — so we never map one onto the other.
If the shipment carries padding…
…the gate removes it before you pay: the parcel reaches the cloud much smaller and the difference drops into your jar.
If it's already lean…
…it passes untouched. There's nothing to remove there — and we don't promise savings where there are none.
Not every shipment carries the same padding: in this run, the measured trim per scenario ranged from -0.2% to 100%. The ones that don't save get published too.
gpt-4o-miniEvery kind of saving, counted separately — so no number gets inflated.
Savings saves in several ways: it avoids whole calls, removes text you didn't need, packs what's sent more tightly, and switches on the provider's discounts. Each way keeps its own ledger with its own measured figure — and we never add them up into one marketing number.
Each ledger is measured and published on its own against the OpenAI API; the ones that do not clear zero do not appear. A percentage without its model next to it means nothing: the same lever can save on one model and cost on another.
avoided_calls
LLM, retrieval and tool calls that never executed at all.
eliminated_tokens
Information dropped because it simply wasn't needed.
compressed_tokens
Information kept, but encoded with fewer tokens.
Run it in your own app. Your prompts never touch our servers.
Drop the SDK in front of your provider. Every response carries a full SavingsReport you can chart — the same numbers the dashboard reads.
# 1 · the package pnpm add @bivelio/savings-layer # 2 · the license (from your dashboard) export BIVELIO_LICENSE_KEY=… # 3 · wrap the call — this is the entire code change const layer = createSavingsLayer({ provider, licenseKey, … }); const { text, report } = await layer.generate({ model, messages }); console.log(report.cost.netSavings) // ← the saving, measured
Where your automations already live. If it calls your provider's API with your key, it gets measured — the SDK wraps the call; the gateway sits in front.
Claude Code on an API key? See its real spend, token by token.
One command puts a local meter in front of your terminal. No code changes, nothing to configure — and your prompts never leave your machine: only counts and cost reach this dashboard.
npm i -g @bivelio/savings-layer export BIVELIO_SERVICE_KEY=bvk_… bvsala claude
The gateway measures every request. With a PRO license it also applies one lever by default: it shortens the cache-write TTL while your session stays hot — that changes the price, never the content the model reads. The other levers stay off until you switch them on. The savings-mode skill is available for your caps, with its effect still to be measured. It also works with Codex CLI, Goose and Qwen Code — and bvsala doctor prints the exact configuration for the rest. See how it works →
The skill that stretches your subscription caps
No invoice doesn't mean no pain: the five-hour window and the weekly cap. savings-mode is a pure-instructions skill — no proxy, your traffic untouched — that cuts the two masses that actually burn cap: model output and tool results. It's free; what you pay Savings for is knowing how much it saves you.
cache reads
The whole context is re-read every turn — which is why keeping it lean matters.
cache writes (one hour)
Every tool result is cache-written at a premium before being re-read.
model output
Under one percent of the tokens; the skill attacks it directly.
Measured with the meter over 959 real Claude Code requests, against the provider's own counters (2026-09-01). The skill's own effect will be published once A/B-measured — never before.
npx -y @bivelio/savings-layer skillOne command: installs into ~/.claude/skills and activates by itself when the conversation turns to limits or saving tokens.
A small fee per seat. And commission only on what you save.
No free tier, no trial: what you can check before paying are the public measurements on this page. The fee is per seat, in EUR, and the commission comes out of the measured saving — never out of your pocket.
PRO — billed per seat. Or €39 / seat / year (save ~19%). Enterprise, with packaged features, comes later.
VAT not included. If your company has an EU VAT number, the reverse charge applies.
A seat is one active machine or service identity, not a person — billed only if active on 3+ days in a month. CI and ephemeral runners don't count.
- Every BV‑SALA optimization, uncapped
- A receipt per request: SavingsReport with four ledgers, measured and estimated kept apart
- Savings dashboards & team analytics
- Your prompts never leave your infrastructure
And how much would you pay us?
EstimatePlay with your numbers: pick provider and model, and see your savings, our extra and what you keep — with the real catalog prices and the real rate.
Loading the price catalog…
How what you pay us is calculated
- Your first month: only €4 per seat. We measure everything and charge no commission — at close you'll see the exact statement of what it would have cost.
- After that, commission only on measured savings: the kind you can reconcile against your provider's invoice. Estimated savings are never billed.
- And you always keep 90% or more of every euro saved: the commission only exists when your measured savings exist.
- Tiered, each bracket at its own rate — 10% ≤ €1,000 · 8% ≤ €5,000 · 6% ≤ €20,000 · 4% +. The more you save, the lower our percentage.
- A month with no savings? Commission: €0. No lock-in — and your first variable invoice only contains numbers you already saw on your statement.
