AssemblyAI Free Credits: $50 Free, No Card (2026)
$50 free credits, no credit card
AssemblyAI gives new accounts $50 in free credits with no credit card required. AssemblyAI's pricing FAQ says this covers up to 185 hours of pre-recorded or 333 hours of streaming transcription, depending on the model. Free accounts can use pre-recorded transcription, Speech Understanding and streaming, but not the LLM Gateway, and have lower concurrency limits until they upgrade.
Who is eligible
- New AssemblyAI account
- No credit card required
How to claim it
- Sign up at assemblyai.com/dashboard/signup
- Copy your API key from the dashboard
- Send audio to the pre-recorded (/v2/transcript) or streaming endpoint; usage is deducted from the $50
Limits
| Credit | $50 (one-time) |
|---|---|
| AssemblyAI's estimate | Up to 185 hours pre-recorded or 333 hours streaming |
| Pre-recorded concurrency (free) | 5 jobs at once (paid: 200+) |
| Streaming (free) | 5 new sessions per minute (paid: 100+) |
| LLM Gateway | Not available on the free tier |
Catches & caveats
AssemblyAI doesn't publish an expiry for the $50. Its hour estimates (185 pre-recorded, 333 streaming) are AssemblyAI's own and depend on the model you choose.
AssemblyAI gives every new account $50 in free credits, and you can start transcribing without a credit card. AssemblyAI's pricing FAQ says the free tier covers up to 185 hours of pre-recorded or 333 hours of streaming transcription. The credit is a one-time balance, not a monthly allowance, and free accounts have lower concurrency limits and no access to the LLM Gateway. When the $50 is used up, you add a payment method and continue on pay-as-you-go.
What the $50 covers
AssemblyAI's support article says free accounts can use:
- Pre-recorded transcription and Speech Understanding features through the async `/v2/transcript` endpoint.
- Streaming speech-to-text through the streaming WebSocket endpoint.
The LLM Gateway, which runs language models over your transcripts, is not available on the free tier. AssemblyAI's support article doesn't list the newer Voice Agent, Sync and Dictation APIs either way, so check your dashboard before relying on them with free credit.
Usage is deducted at the normal pay-as-you-go rates. These are the main rates on AssemblyAI's pricing page in October 2026, with what $50 buys at each:
| Model or API | Rate | What $50 buys (approx.) |
|---|---|---|
| Universal-3.5 Pro (pre-recorded) | $0.21 / hour | ~238 hours |
| Universal-2 (pre-recorded, 99 languages) | $0.15 / hour | ~333 hours |
| Universal-3.6 Pro Realtime (streaming) | $0.45 / hour | ~111 hours |
| Universal-Streaming (English or multilingual) | $0.15 / hour | ~333 hours |
| Sync API (clips up to 2 minutes) | $0.30 / hour | ~167 hours |
| Dictation API | $0.62 / hour | ~81 hours |
| Voice Agent API | $4.50 / hour ($0.075/min) | ~11 hours |
The right-hand column is our arithmetic: $50 divided by the hourly rate, before add-ons. AssemblyAI's own FAQ figure of 185 pre-recorded hours is lower than the 238 hours that $50 ÷ $0.21 gives for Universal-3.5 Pro; the page doesn't explain the assumption behind it, so treat both as estimates.
Add-ons use the same credit
Many features are billed per hour on top of the base model. On pre-recorded audio with Universal-3.5 Pro, AssemblyAI lists keyterms prompting at $0.05/hr, prompting at $0.05/hr, speaker diarization at $0.02/hr and medical mode at $0.15/hr. Speech Understanding features are separate line items: translation $0.06/hr, custom formatting $0.03/hr, entity detection $0.08/hr, sentiment analysis $0.02/hr, key phrases $0.01/hr, topic detection $0.15/hr, and speaker identification or summarization from $0.02/hr at low effort. Guardrails such as PII text redaction ($0.08/hr) and content moderation ($0.15/hr) are also extra.
A worked example: Universal-3.5 Pro with diarization, entity detection and PII text redaction costs $0.21 + $0.02 + $0.08 + $0.08 = $0.39 per hour, so $50 covers about 128 hours instead of 238.
Who qualifies
AssemblyAI's pricing page says you can create an account and start transcribing immediately with no credit card. The support article describes the free tier as a limited trial: adding a credit card at any time upgrades you out of it. AssemblyAI doesn't publish other conditions such as country, business status or age on these pages.
How to claim it
- Create an account at assemblyai.com/dashboard/signup.
- Copy the API key shown in your dashboard.
- Submit audio to the pre-recorded endpoint or open a streaming session. Charges come off the $50 automatically.
- Keep an eye on your balance on the Billing page. AssemblyAI suggests enabling billing alerts so you know before the credit runs out.
import assemblyai as aai
aai.settings.api_key = "YOUR_API_KEY"
config = aai.TranscriptionConfig(speaker_labels=True)
transcript = aai.Transcriber().transcribe("https://example.com/interview.mp3", config)
print(transcript.text)If you don't set a model, the API uses its default. Specify the model in the request when you want to control which hourly rate you pay.
Which model to test with your credit
AssemblyAI's lineup has grown, and the model you choose decides how far the $50 goes. Here is how AssemblyAI describes each one on its pricing page:
| Model or API | Type | What AssemblyAI says it is for |
|---|---|---|
| Universal-3.5 Pro | Pre-recorded | Its most accurate async model; 18 languages, native code switching and its best speaker diarization |
| Universal-2 | Pre-recorded | Lower-priced model trained on over 12.5 million hours of audio; 99 languages |
| Universal-3.6 Pro Realtime | Streaming | Most accurate realtime model for voice agents; 32 languages, context carryover, voice isolation |
| Universal-Streaming | Streaming | Low-latency English transcription for production voice apps |
| Universal-Streaming Multilingual | Streaming | English, Spanish, German, French, Portuguese and Italian at the same price |
| Sync API | Single request | Universal-3.5 Pro accuracy for clips up to 2 minutes, with results in about 134 ms at the median |
| Dictation API | Single request | Cleaned-up text from one spoken utterance (up to 120 seconds), 19 languages |
A practical split: test accuracy on your own recordings with both Universal-3.5 Pro and Universal-2, since the price difference ($0.21 vs $0.15 per hour) adds up at volume. If you are building something live, compare Universal-Streaming with Universal-3.6 Pro Realtime; the realtime Pro model costs three times as much per hour ($0.45 vs $0.15).
Real uses for the free credit
- Transcribing a podcast back catalogue: 150 one-hour episodes on Universal-2 cost 150 × $0.15 = $22.50, under half the credit.
- Evaluating meeting-notes accuracy: run 50 hours of real meetings through Universal-3.5 Pro with speaker diarization: 50 × $0.23 = $11.50.
- Prototyping live captions: 100 hours of Universal-Streaming sessions cost $15, but remember you can open only 5 new sessions a minute on the free tier.
- Trying the Voice Agent API: at $4.50 an hour, $50 is about 11 hours of connected conversation, enough for a few hundred short test calls if the API is enabled for your account.
These are worked examples using the published rates, not AssemblyAI figures.
Free-tier limits
| Limit | Free account | Paid (pay-as-you-go) |
|---|---|---|
| Pre-recorded jobs processed at once | 5 | 200+ |
| New streaming sessions per minute | 5 | 100+, rising 10% automatically when you use 70% of the limit |
| LLM Gateway | Not available | Available, priced per million tokens by model |
| Total concurrent streaming sessions | No hard cap stated | No hard cap stated |
Pre-recorded jobs over the limit are not rejected. AssemblyAI's concurrency docs say they go into a queue and are processed first-in, first-out as earlier jobs finish, and you may get a throttling email (at most once a day). If your balance falls below zero, your concurrency limit drops to 1.
For batch work on the free tier, queueing is usually fine: you can submit a large backlog and it will work through it five jobs at a time. For live apps, the limit of 5 new streaming sessions per minute is the constraint to plan around.
When the credit runs out
AssemblyAI's support article explains what happens:
- Using the API with a negative balance returns an error asking you to top up.
- To restore access, add a payment method, add funds or enable autopay. Access resumes once the balance is positive.
- If you upgrade before using the full $50, the unused amount stays on your account.
Paid usage is billed monthly on actual usage, with no minimum commitment, upfront fees or contracts on pay-as-you-go. Invoices are generated at the start of each month for the previous month. Volume discounts and custom concurrency are available through sales.
Worked costs after the free credit
- A meeting-notes tool transcribing 500 hours a month with Universal-3.5 Pro and diarization: 500 × ($0.21 + $0.02) = $115 a month.
- A call-center archive of 2,000 hours a month on Universal-2: 2,000 × $0.15 = $300 a month.
- Stereo recordings: multichannel audio is billed per channel, so a 1-hour stereo file transcribed per channel counts as 2 hours.
- Streaming: billed on connection open time, not on audio sent, so idle sessions cost money.
Getting the most out of $50
- Test with Universal-2 first if your audio is clean or you need one of its 99 languages; it is the cheapest pre-recorded model at $0.15/hr.
- Switch on add-ons only for the files that need them. Each one is billed per hour of audio.
- Close streaming connections promptly. Streaming is billed on session time.
- Use the Playground on AssemblyAI's site for quick comparisons before writing code.
- Pair transcripts with a free LLM. Since the LLM Gateway isn't on the free tier, you can summarise transcripts with a free model API such as the Gemini API free tier or Groq while testing.
How it compares with other free speech-to-text options
| Provider | Free offer | Notes |
|---|---|---|
| AssemblyAI | $50 one-time, no card | Up to 185 hrs pre-recorded per AssemblyAI; 5 concurrent jobs on free |
| Deepgram | $200 credit, no card | Larger credit; also covers TTS and voice agents |
| ElevenLabs free plan | 10,000 credits a month | Scribe speech-to-text plus TTS; non-commercial |
| Groq free tier | Rate-limited free plan | Whisper models with daily limits |
| Hugging Face Inference | No monthly credits on free accounts; $2/month on PRO | Open-source speech models |
For a broader view, see our guide to free AI API keys and how to get free AI credits.
Billing details worth knowing
- Pre-recorded audio is billed per hour of audio processed, and multichannel files are billed per channel.
- Streaming is billed on session time: AssemblyAI says it charges for how long the connection is open, not for how much audio you send.
- The Voice Agent API is billed per second of connected conversation time, with speech-to-text, the agent LLM, text-to-speech and hosting all included in the $4.50 hourly rate.
- LLM Gateway prices are per million tokens and vary by model. AssemblyAI notes that in-region (US or EU) routing costs 10% more than the global rates it lists.
- You can monitor spend on the Billing page of the dashboard, where AssemblyAI suggests setting billing alerts and, once you have a payment method, autopay.
Common mistakes
- Expecting the credit to renew monthly. It is a one-time $50.
- Planning a product around the LLM Gateway on free credit. It isn't available until you upgrade.
- Counting stereo audio as one hour. Multichannel files are billed per channel.
- Letting the balance go negative. API calls fail and concurrency drops to 1 until you top up.
- Confusing marketplace offers with the direct offer. AssemblyAI is also sold through AWS Marketplace, where billing goes through your AWS account; the $50 direct sign-up credit described here is the one on AssemblyAI's own site.
Frequently asked questions
›Is AssemblyAI free?
New accounts get $50 in free credits with no credit card required. After that you pay per hour of audio on the pay-as-you-go plan.
›How many hours of transcription is AssemblyAI's free credit?
AssemblyAI's pricing FAQ says the free tier covers up to 185 hours of pre-recorded or up to 333 hours of streaming transcription. The exact number depends on the model and add-ons; at Universal-2's $0.15 per hour, $50 is about 333 hours, and at Universal-3.5 Pro's $0.21 per hour it is about 238 hours.
›Do AssemblyAI free credits expire?
AssemblyAI's pricing page and its support article on free signup don't state an expiry.
›Is the $50 monthly?
No. AssemblyAI describes it as $50 of audio processing for free accounts; once it is used, you must upgrade to keep using the API.
›What can free accounts use?
Pre-recorded transcription and Speech Understanding through the async /v2/transcript endpoint, and streaming speech-to-text over WebSocket. The LLM Gateway is not available on the free tier.
›What are the free-tier limits?
Free accounts can run 5 pre-recorded transcriptions at once (paid accounts start at 200) and open 5 new streaming sessions per minute (paid accounts start at 100). Extra pre-recorded jobs are queued rather than rejected.
›What happens to unused credit if I upgrade?
AssemblyAI says that if you upgrade before using the full $50, the unused amount stays on your account.
›What happens when the credit runs out?
Requests made with a negative balance return an error asking you to top up. Add a payment method and funds, or enable autopay, and access resumes once the balance is positive.
Sources
Checked against these pages on Oct 17, 2026. Providers change offers often: always confirm on the official page before you rely on it.
- Official: AssemblyAI pricing(www.assemblyai.com)
- Official: AssemblyAI support, Can I sign up for free?(support.assemblyai.com)
- Official: AssemblyAI docs, pre-recorded concurrency(www.assemblyai.com)
- Official: AssemblyAI docs, streaming concurrency(assemblyai.com)
Similar deals
Google Cloud Free Trial
$300 Welcome credit valid for 90 days
New Google Cloud customers get a $300 Welcome credit to spend over a 90-day Free Trial on most Google Cloud services. You must add a valid payment method, but you are not billed during the trial. The credit cannot pay for the Gemini API in AI Studio or for partner models offered as managed APIs, and trial accounts cannot add GPUs, use Marketplace or request quota increases. If you don't upgrade, resources are stopped at the end and deleted after a 30-day grace period. Separately, the always-free Free Tier keeps running with no end date.
Deepgram Speech API
$200 free credit for new accounts, no credit card
Every new Deepgram account gets $200 in free credit on the Pay As You Go plan, with no credit card required. Deepgram says this is about 43,000 minutes (over 700 hours) of Nova transcription, and the credit works across speech-to-text, text-to-speech, the Voice Agent API and audio intelligence. Deepgram's pricing page says the credit lasts until used, but its developer docs say free promotional credits expire one year from signup.
Gemini API / Google AI Studio
Free tier: Gemini 3.8 Flash, 3.6 Flash, 3.5 Flash-Lite, 2.5 Pro, 2.5 Flash, Gemma 4 and more, rate-limited
Google's Gemini API has a Free tier where many models are listed as 'Free of charge' for input and output, including Gemini 3.8 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini Embedding 2 and Gemma 4. Newer flagship, image, video and music models (Gemini 4 Argon, Gemini 3.1 Pro Preview, Nano Banana, Veo 3.1, Lyria) are paid-only. Free-tier prompts and responses may be used to improve Google's products and can be read by human reviewers. Rate limits apply per project and are shown in Google AI Studio rather than published as a fixed table.
GroqCloud API
Free plan: rate-limited access to models like gpt-oss-120b, Qwen3.8-27B and Whisper
GroqCloud has a Free plan with per-model rate limits published on its rate-limits docs page, for models including gpt-oss-120b, gpt-oss-20b, Qwen3.8-27B, Whisper and Orpheus text-to-speech. The paid Developer plan is pay-as-you-go with much higher limits (for example 1K RPM instead of 30 for gpt-oss-120b) plus Batch and Flex processing. Limits apply at the organization level and your exact numbers are shown on the console Limits page.