LMNT pricing: should I use Indie, Pro, or Premium for a real-time voice agent, and how do overages work per 1K characters?

Most teams evaluating LMNT for a real-time voice agent end up with the same two questions: which paid plan (Indie, Pro, or Premium) actually fits my usage pattern, and what will I really pay once I cross the included characters and start incurring overages per 1K characters?

Quick Answer: Use Indie if you’re validating a single agent with light traffic, Pro if you’re running a serious pilot or early production with steady users, and Premium if you’re building a core product feature or multi-agent fleet that needs predictable, high-volume economics. Overage pricing is calculated per 1,000 characters on top of your included monthly characters, so your real cost scales directly with how much speech you generate—not with concurrent users or sessions.

Why This Matters

For a real-time voice agent, text-to-speech isn’t a side cost; it’s the meter that runs on every conversation. If you pick a plan that’s too small, you’ll burn through characters and get surprised by overages. If you go too big, you sit on unused capacity. The right LMNT tier lets you:

Hit conversational latencies (150–200ms streaming) without worrying about concurrency limits.
Clone as many voices as you need from minimal audio (5-second samples) without separate SKU drama.
Keep your cost per 1K characters predictable as you scale from prototype to production.

Key Benefits:

Predictable scaling: Character-based pricing and no concurrency or rate limits means your costs track usage cleanly, even as you add more users or agents.
Production-grade performance: All tiers share the same core engine—150–200ms low-latency streaming and 24 languages—so you don’t trade quality for price.
Builder-friendly path: Start with Indie or even the free Playground, then step up to Pro or Premium when your real-time voice agent becomes a revenue path, not just an experiment.

Core Concepts & Key Points

Concept	Definition	Why it's important
Characters vs. sessions	LMNT bills on characters (text) converted to speech, not on calls or seats.	Lets you estimate cost directly from your prompt lengths and expected talk time.
Included vs. overage	Each plan includes a monthly character allowance; overages are charged per additional 1K characters.	You can “burst” past your plan without hitting throttles—just pay for the extra usage.
No concurrency limits	LMNT doesn’t cap concurrent streams or requests across paid tiers.	You can handle spikes in traffic (launch days, events, game peaks) without worrying about rate-limit errors.

Note: Exact prices, included characters, and overage rates can change over time. Always confirm the latest numbers on LMNT’s pricing page before you commit. The guidance here focuses on how to choose a tier and think about overages for real-time agents.

How It Works (Step-by-Step)

At a high level, your real-time voice agent’s LMNT cost comes down to:

Estimate characters per minute of speech.
As a rough rule of thumb, 1 second of natural speech ≈ 12–18 characters (including spaces and punctuation), so 1 minute is in the ballpark of 800–1,200 characters depending on how dense your text is. For most agents, 1K characters ≈ 1 minute of output audio is a safe planning number.
Map your traffic to a monthly character budget.
Take your expected average minutes of speech per conversation × number of conversations per month. Then convert those minutes to characters using the 1K chars ≈ 1 minute guideline.
Pick the smallest plan that covers your normal load and price the overages.
Choose Indie, Pro, or Premium based on your baseline usage; assume some buffer. Then look at LMNT’s per-1K character overage rate and decide whether your likely overage volume makes the higher tier more cost-effective.

Because LMNT has no concurrency or rate limits across plans, you’re not choosing a tier to unlock more “slots” for agents—you’re choosing a tier to optimize your cost per 1K characters given how much speech your users will actually consume.

1. When Indie Makes Sense

Indie is typically the right starting point if:

You’re building a single voice agent or a small set of agents.
You have low to moderate traffic (e.g., internal tool, early beta, or a niche game).
You want production-ready streaming latency (150–200ms) and voice cloning, but you’re still validating fit and usage patterns.

Indie is also a good fit if:

You’re integrating via the API after playing in the free Playground.
You expect to run in the tens to low hundreds of hours of TTS per month, not thousands.
Overages are expected to be small relative to your included characters.

How to sanity-check Indie for your agent:

Estimate your monthly minutes of TTS:
- 100 users × 10 conversations/month × 2 minutes of agent speech = 2,000 minutes.
Convert to characters:
- 2,000 minutes × ~1K characters/minute ≈ 2M characters.
Compare to Indie’s character allowance:
- If Indie’s included characters (check the pricing page) cover 2M with some headroom, you’re safe.
- If you’re immediately using multiples of Indie’s allowance, you’re better off on Pro to avoid paying most of your usage as overage.

2. When Pro Is the Better Default

Pro is usually the “real product” default for voice agents that have graduated from experiments:

You’re running a public app, agent, or game with real users.
You see steady or growing traffic and run marketing campaigns.
You expect ongoing iteration, multiple environments (staging, prod), and more cloned voices.

Choose Pro when:

Your projected monthly characters are more than the Indie allowance but still not at large-enterprise scale.
You want a more favorable cost per 1K characters on both included usage and overages.
You’re coordinating with a small team across multiple projects or agents.

Pro is also the tier where overage economics typically start to matter. If your predicted usage sits near the top of Indie every month, the overage rate might make Pro cheaper overall—even if the headline price looks higher.

3. When Premium Is the Right Call

Premium is for teams where voice is a core product surface, not an experiment:

You’re running multiple agents, tenants, or game shards, all speaking concurrently.
You expect thousands of hours per month of streaming TTS.
Your users are global, and you’re leaning on 24 languages and code-switching (mid-sentence language changes).

You’re in Premium territory if:

Your monthly characters are at or above the Pro plan’s capacity.
Overages on Pro would make your effective cost per 1K characters worse than Premium’s included rate.
You want to lock in stable economics ahead of a product launch or seasonal spike.

Premium also makes sense if you:

Need to guarantee no concurrency bottlenecks during heavy peaks.
Plan to run many voice clones (e.g., per-tenant or per-character voices) and want to treat them as standard usage, not custom add-ons.
Are preparing for enterprise conversations around security (SOC-2 Type II) and compliance.

How Overages Work Per 1K Characters

Overage charging is straightforward:

Each plan has a monthly character allowance.
Once your account’s TTS usage passes that allowance in a given billing period, you start accruing overages.
Overage is billed per additional 1,000 characters.
LMNT tracks total characters generated beyond your allowance and multiplies by the per-1K overage rate for your tier.
All usage is summed across agents, environments, and voices.
If you run multiple agents or apps under one LMNT account (or workspace), their combined usage determines when you hit the allowance and how much overage you pay.
No penalties for spikes beyond capacity.
Because there are no concurrency or rate limits, crossing your included characters doesn’t throttle your real-time voice agent. You keep streaming; you just pay the incremental per-1K character cost.

Always check LMNT’s pricing page for the current per-1,000 character overage rate for Indie, Pro, and Premium. The tier you pick affects this rate, so your projected overage volume can influence which plan is cheapest overall.

Quick Planner: Estimating Overages

To make this concrete, use this simple mental model:

1K characters ≈ 1 minute of speech (typical agent speaking rate).
Overage cost per minute ≈ overage price per 1K characters.

Example workflow:

For each conversation, estimate agent talk time:
- 3 minutes of agent speech → ~3K characters.
Multiply by conversations per month:
- 5,000 conversations × 3K characters ≈ 15M characters.
Compare 15M to your plan:
- If Pro includes, say, 10M characters and you generate 15M, you’d pay overage on ~5M characters.
- 5M / 1K = 5,000 overage units × (Pro overage rate per 1K characters).

If you find that:

Overages on Indie would be most of your usage, or
Overages on Pro are consistently high,

then upgrading to the next tier usually lowers your effective cost per 1K characters and simplifies budgeting.

Common Mistakes to Avoid

Underestimating how “chatty” your agent is:
Many teams size plans based on request counts, not speech length. If your agent’s answers are long-form (tutors, storytellers, broadcasters), your characters per session will skew high.
Avoid it by: Logging output text length (characters) in development and extrapolating from real data, not guesses.
Ignoring overage economics when picking a tier:
Looking only at the base monthly price without modeling overage volume can make Indie look cheaper when Pro or Premium would actually reduce your effective price per 1K characters.
Avoid it by: Calculating projected character usage, then modeling cost across Indie, Pro, and Premium using each tier’s overage rates.

Real-World Example

You’re launching a real-time customer support voice agent that handles common questions and escalates complex issues to a human.

Your product assumptions:

~800 active users in month one.
Each user has ~5 voice sessions/month.
The agent speaks about 90 seconds per session on average.

Let’s walk it:

Estimate characters per session:
- 90 seconds ≈ 1.5 minutes.
- 1.5 minutes × ~1K characters/minute ≈ 1,500 characters.
Total characters per month:
- 800 users × 5 sessions/user = 4,000 sessions.
- 4,000 sessions × 1,500 characters ≈ 6,000,000 characters/month.
Choosing a plan:
- If Indie’s allowance is significantly below 6M characters, you’d pay substantial overages.
- Pro likely covers that 6M with room for growth and offers better overage pricing if you temporarily spike to 8–10M characters.

In this scenario, even if Indie’s base subscription looks cheaper, the overage math can easily make Pro the better economic choice for a real-time agent that’s already tied to a production workflow.

Pro Tip: Before you commit to a tier, prototype your agent using the LMNT Playground and a small API integration, then log output character counts for a week of realistic traffic or scripted load tests. Use that real data to run an Indie vs. Pro vs. Premium cost comparison—including overages—so your first bill matches your mental model.

Summary

For real-time voice agents, LMNT’s pricing is all about characters:

Indie is ideal for early builds, low-traffic agents, and indie-scale games where you want production-grade latency without committing to high volume.
Pro fits most teams running serious pilots or early production—steady users, multiple agents, and enough traffic that overage economics matter.
Premium is for high-volume, multi-agent or multi-tenant products where TTS is core to the experience and you want predictable, scalable character pricing.

Across all tiers, you get the same core engine—150–200ms low-latency streaming, 24 languages, studio-quality cloning from a 5-second recording, and no concurrency or rate limits. Overages are calculated per 1K characters once you pass your included monthly characters, so your costs scale cleanly with how much your agents actually talk.

Next Step

Get Started