Yes — But It Expires After 12 Months
Amazon Polly has a free tier that gives you 5 million Standard characters and 1 million Neural characters per month. That's roughly 5 hours of Standard audio or 1 hour of Neural audio for free, every month, for 12 months from your first API call. After that, you pay per character — and the pricing varies wildly depending on which engine you use.
The part that trips people up: the free tier is time-limited, not permanent. The 12-month clock starts when you first use Polly, not when you create your AWS account. And the newer Generative and Long-Form engines have their own (smaller) free allocations. Here's exactly what you get and what happens when it runs out.
Amazon Polly Free Tier Breakdown (June 2026)
| Engine | Free Per Month | Duration | After Free Tier |
|---|---|---|---|
| Standard | 5M characters | 12 months | $4/1M chars |
| Neural | 1M characters | 12 months | $16/1M chars |
| Long-Form | 500K characters | 12 months | $100/1M chars |
| Generative | 100K characters | 12 months | $30/1M chars |
As of July 2025, new AWS customers also receive up to $200 in AWS Free Tier credits applicable across services, including Polly. These credits are available for 6 months after account creation and stack on top of Polly's own free tier. For the full pricing breakdown across all four engines, see our Amazon Polly pricing guide.
How Much Audio Is 5 Million Characters?
Abstract character counts are useless without context. Here's what 5 million Standard characters actually translates to:
- ~5 hours of audio per month (Standard engine, average speaking rate)
- ~3,300 pages of text (at ~1,500 characters per page)
- ~33 medium blog posts (at ~150K characters each)
- ~10 podcast episodes worth of narration (30 min each)
For the Neural engine, 1 million characters gives you roughly 1 hour of audio per month. That's enough for occasional use — a few blog posts or a short video narration — but you'll run out quickly if you're doing bulk audio generation.
The Generative engine's 100K character free allocation? That's about 6 minutes of audio. Basically a demo, not a usable free tier. And at $30/1M characters after that, Generative is the most expensive Polly engine. Use our TTS cost calculator to estimate costs at your specific volume.
What Happens After 12 Months?
The free tier expires — no grace period, no warning unless you've set up billing alerts. After 12 months, every character you synthesize costs money:
Post-Free-Tier Pricing
The Standard engine at $4/1M is still the cheapest managed TTS API on the market — cheaper than OpenAI TTS ($15/1M), Fish Audio ($15/1M), and ElevenLabs ($180–$300/1M depending on plan). But the voice quality reflects the price. For a full comparison across 11 services, check our TTS pricing comparison.
How to Avoid Surprise AWS Bills
This is the biggest gotcha with Polly's free tier. If you exceed your monthly allocation or your 12 months expire, billing starts immediately with no notification unless you've configured alerts. I've seen developers on Reddit discover $50–$200 surprise charges because they forgot the free tier was time-limited.
Set these up before making your first API call:
- AWS Budgets — Create a zero-spend budget with email alerts. Go to AWS Billing → Budgets → Create Budget → set a $1 monthly threshold. You'll get an email the moment any charge appears.
- CloudWatch billing alarm — More granular than Budgets. Set up an alarm for EstimatedCharges > $0 in the us-east-1 region. Triggers within hours of any charge.
- Calendar reminder — Mark your 12-month expiry date. Obvious, but most people forget.
Free Tier Fine Print Most Guides Skip
- Per-account, not per-project — The free tier applies to your entire AWS account, not individual projects or applications. If you have three apps using Polly, they share the same 5M Standard / 1M Neural allocation.
- SSML tags count as characters — If you use Speech Synthesis Markup Language for pronunciation or emphasis, those tags consume your character quota. A sentence with SSML can be 2–3x the character count of plain text.
- Speech Marks requests count too — If you request Speech Marks (word timing metadata) alongside audio, that counts as a separate character consumption. Requesting both audio and Speech Marks for the same text doubles your usage.
- Region doesn't matter — Free tier is the same regardless of AWS region. Pricing after the free tier is also region-independent for Polly.
- Commercial use is fine — Unlike some free tiers, Polly's free tier has no restriction on commercial use. You can use the generated audio in products, apps, and content from day one.
Amazon Polly Free Tier vs Other Free TTS Options
Polly has the most generous managed free tier, but it's not the only option. Here's how it stacks up:
| Service | Free Allocation | Expires? | Voice Quality |
|---|---|---|---|
| Amazon Polly | 5M chars/mo (Standard) | Yes, 12 months | 6/10 (Standard), 7/10 (Neural) |
| ElevenLabs | 10K chars/mo | No, permanent | 9/10 |
| Google Cloud TTS | 1M chars/mo (Standard) | No, ongoing | 7/10 |
| Azure TTS | 500K chars/mo | No, ongoing | 7/10 |
| Murf AI | Limited, watermarked | No, permanent | 7/10 |
| OpenAI TTS | None (pay-per-use only) | N/A | 8/10 |
| Chatterbox | Unlimited (self-hosted) | Never | 9/10 |
| Kokoro | Unlimited (self-hosted) | Never | 8/10 |
The takeaway: Polly's free tier is the biggest among managed cloud APIs, but it's the only one with an expiration date. Google Cloud's 1M chars/month is smaller but never expires. ElevenLabs' 10K chars/month is tiny but includes the best voice quality. And if you're willing to self-host, open-source models like Kokoro and Chatterbox give you unlimited characters forever.
Who Should Use the Polly Free Tier (and Who Shouldn't)
Good Fit
- AWS developers adding TTS to existing apps
- Prototyping and proof-of-concept projects
- Low-volume production use (under 5M chars/mo)
- IVR, notifications, and accessibility features
- Teams who need SSML control over pronunciation
Not a Good Fit
- Content creators who need natural-sounding narration
- Anyone who needs voice cloning (Polly doesn't offer it)
- High-volume use after the free tier expires
- Users who want a web interface (Polly is API-only)
- Anyone not already on AWS infrastructure
How to Start Using Amazon Polly Free
- Create an AWS account — You'll need a credit card on file even for the free tier (standard AWS requirement). You won't be charged unless you exceed the free allocation.
- Set up billing alerts first — Before using Polly, go to AWS Budgets and create a $0 threshold alert. This protects you from surprise charges.
- Create an IAM user with Polly permissions — Don't use your root account for API calls. Create a dedicated IAM user with the AmazonPollyFullAccess policy.
- Test in the AWS Console first — The Polly Console lets you type text and hear voice samples without writing any code. Test different voices and engines before committing to an integration.
- Integrate via SDK — AWS SDKs for Python (boto3), JavaScript, Go, Java, and .NET all support Polly. A basic synthesis call is 5–10 lines of code.
For a more detailed walkthrough and SDK examples, check our Amazon Polly service page.
The Bottom Line
Amazon Polly's free tier is genuinely useful — 5M Standard characters per month is enough for most development and light production use. But it's not "free forever." Mark your calendar for 12 months out, set up billing alerts on day one, and know that the cheapest engine (Standard at $4/1M) still has the voice quality to match its price.
If you need permanently free TTS with better voice quality, look at open-source models — Kokoro and Chatterbox cost $0 forever. If you need a managed API without expiration, Google Cloud TTS offers 1M chars/month ongoing. For the full picture of every TTS pricing option, see our TTS pricing comparison or estimate your costs with our calculator.
Related Guides
By TextToLab Research Team · Last verified June 2026. Free tier details from AWS Polly pricing page. Pricing confirmed against official documentation. Voice quality ratings from our independent testing. ElevenLabs affiliate link disclosed — all other recommendations are independent.