Imported from jeremylongshore/tons-of-skills-marketplace (skills/.curated/elevenlabs-cost-tuning/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill elevenlabs-cost-tuning. Copyright stays with the author (MIT).

ElevenLabs Cost Tuning

Overview

Optimize ElevenLabs costs through model selection (Flash = 50% savings), character-efficient text processing, audio caching, and real-time quota monitoring. ElevenLabs bills by character for TTS and by audio minute for STT.

Prerequisites

• ElevenLabs account with usage dashboard access
• Understanding of your monthly character consumption
• Access to billing at https://elevenlabs.io/app/subscription

Instructions

Step 1: Understand the Billing Model

TTS billing (by character):

Other feature billing:

Plan character limits:

Steps 2–6: Apply the cost levers

Work through the levers in order of savings-per-effort. Each ships as a small, drop-in TypeScript helper — the full source for every step is in implementation.md.

1. Model-based reduction — route each request through selectCostEffectiveModel() so functional audio (greetings, notifications) uses Flash/Turbo at 0.5x while premium, customer-facing output keeps full-quality models. Biggest single win (50%).
2. Character-efficient text — run copy through optimizeTextForTTS() to strip markdown, HTML, and redundant whitespace/punctuation before billing counts it (5–15%).
3. Real-time quota monitoring — getQuotaStatus() returns used/remaining/percent, a per-day budget until reset, and a projectedOverage flag from the current usage rate.
4. Cost-aware request guard — guardedTTS() refuses a call that exceeds remaining quota and force-downgrades to Flash above 90% usage, preventing hard overages.
5. Usage tracking — trackUsage() + getUsageSummary() roll up credits by model and operation and compute a cache-hit rate so you can see where spend actually goes.

Minimal skeleton — the guard is the piece most workloads adopt first:

import { guardedTTS } from "./elevenlabs/cost-aware-tts";

// Notifications auto-route to Flash (0.5x); guard blocks or downgrades near the limit.
const stream = await guardedTTS("Your table is ready.", VOICE_ID, "notification");

Cost Optimization Checklist

Output

Applying this skill produces:

• A cost-aware TTS layer — selectCostEffectiveModel() + guardedTTS() that pick the cheapest acceptable model per content type and refuse/downgrade calls near the quota.
• A text optimizer — optimizeTextForTTS() returning { optimized, originalLength, savedCharacters }.
• A live quota picture — getQuotaStatus() returning plan, used, limit, remaining, pctUsed, dailyBudget, and a projectedOverage boolean.
• A usage roll-up — getUsageSummary() reporting total credits/characters, spend by model and operation, and cache-hit rate over a trailing window.

Together these turn an unmonitored, single-model TTS integration into one with per-request cost control, overage prevention, and a spend audit trail.

Examples

Quick shape (full, runnable scenarios in examples.md):

import { getQuotaStatus } from "./elevenlabs/quota-monitor";

const q = await getQuotaStatus();
console.log(`${q.plan}: ${q.pctUsed}% used, ${q.remaining.toLocaleString()} chars left`);
if (q.projectedOverage) console.warn("On pace to exceed quota this cycle");

• Check quota before a batch run — abort early if the batch would exceed remaining chars.
• Route content to the cheapest acceptable model — Flash for notifications, eleven_v3 for premium.
• Trim characters before billing counts them — strip markdown/HTML with optimizeTextForTTS().
• Roll up 30-day spend — see credits by model and cache-hit rate with getUsageSummary().

See examples.md for the complete code of each.

Error Handling

Resources

• Full implementation walkthrough — Steps 2–6 source
• Worked examples — end-to-end cost scenarios
• ElevenLabs Pricing
• Usage Dashboard
• Subscription Settings
• Models Overview
• For architecture patterns, see the elevenlabs-reference-architecture skill.