Short answer: for recorded audio, AssemblyAI's Universal-2 has the lowest dollar list rate in this comparison at $0.15 per audio hour, or $0.17 with speaker labels, which puts 1,000 hours at $170 a month (prices checked October 28, 2026). Its Universal-Streaming model is also the cheapest live rate, at the same price. Gladia costs less at volume, but in euros. Azure, Google's dynamic batch and OpenAI's gpt-4o-mini-transcribe sit at $0.18 an hour for files, and Amazon Transcribe at $0.36.
Per-hour list rates for recorded files and live audio on ten transcription APIs, plus speaker-label add-ons and three monthly bills worked out.
A speech-to-text API is billed on audio duration, and the headline rate hides three choices that move the bill more than the vendor does: recorded files or live audio, the model tier, and whether speaker labels cost extra. Live transcription costs more than file transcription at every vendor below except AssemblyAI, and more than five times as much on Azure, so a team that can wait a few minutes for a transcript should price batch first.
This guide prices ten APIs on the same terms. Every rate is the vendor's pay-as-you-go list price per audio hour, converted from per-minute where the vendor prints minutes (our arithmetic: per-minute rate times 60). Free credits and negotiated discounts are left out of the bills, because they end or depend on a contract.
Toolradar data: Toolradar's transcription category lists 179 tools, and 69% of them offer a free or freemium plan, so a free allowance is common in the category and says little about the price at volume. The APIs below are the developer end of that list; for apps with an editor and a monthly seat, see the cheapest transcription software and the best AI transcription tools.
Methodology: we read each vendor's own pricing page on October 28, 2026, took the cheapest general-purpose model for the batch scenario, and computed three monthly bills: 50, 1,000 and 20,000 audio hours. Ten APIs from the 179 tools in the category were considered because each sells a self-serve transcription API with published rates. Accuracy, latency and language claims on this page are the vendors' own; Toolradar did not run audio through these APIs. How we rank is explained on our methodology page.
Top Picks
Picked by editorial review, informed by G2 and Capterra review volume and rating and by media mentions, the signals behind our category rankings. How we rate
The lowest dollar rates here for both recorded files and live audio, with every add-on priced per hour.
Starting price
$0.15 /hr files or streams
4.6 (126 reviews)
Best for
Teams that want the cheapest US-dollar rate for files and streams and can price add-ons one by one.
Async rates that fall automatically with monthly volume, with diarization included on files, billed in euros.
Starting price
From €0.16/hr files, priced in euros
4.8 (23 reviews)
Best for
High-volume file transcription where an automatic volume discount beats a negotiated contract.
A low batch rate on the Microsoft invoice, with commitment tiers for steady volume.
Starting price
$0.18 /hr batch, $1/hr real-time
Best for
Microsoft-billed teams transcribing recorded audio in batch jobs.
| Tool | Starting price | Rating | Best for |
|---|---|---|---|
| AssemblyAI | $0.15/hr files or streams | 4.6 126 reviews | Teams that want the cheapest US-dollar rate for files and streams and can price add-ons one by one. |
| Gladia | From €0.16/hr files, priced in euros | 4.8 23 reviews | High-volume file transcription where an automatic volume discount beats a negotiated contract. |
| Azure AI Speech | $0.18/hr batch, $1/hr real-time | n/a | Microsoft-billed teams transcribing recorded audio in batch jobs. |
| Google Cloud Speech-to-Text | $0.003/min dynamic batch | 4.6 556 reviews | Google Cloud customers with file backlogs that can wait for low-urgency processing. |
| OpenAI transcription API | From $0.003/min (estimated) | 4.3 52 reviews | Teams already on the OpenAI API who want transcription on the same key and bill. |
| Rev AI | $0.20/hr Reverb English | 4.7 563 reviews | English-first workloads that may need an occasional human transcript at a known per-minute price. |
| ElevenLabs Scribe | $0.22/hr files, $0.39/hr live | 4.6 1,236 reviews | Voice products already using ElevenLabs speech synthesis that want one vendor for both directions. |
| Speechmatics | $0.24/hr list, Melia 1 files | 4.6 71 reviews | Multilingual or mixed-language audio where included diarization and code-switching matter more than the base rate. |
| Amazon Transcribe | $0.006/min batch, US East | 4.4 17,730 reviews | AWS-native teams that want transcription inside their account, IAM and invoice. |
| Deepgram | Check the current rate on Deepgram | 4.6 475 reviews | Voice-agent builders who need turn detection and will compare a quote against the rates above. |
AssemblyAI
The lowest dollar rates here for both recorded files and live audio, with every add-on priced per hour.
- 4.6 on G2 (126 reviews)
- Starting price
- $0.15/hr files or streams
- Free plan
- No
- Rating
- 4.6 (126 reviews)
Best for: Teams that want the cheapest US-dollar rate for files and streams and can price add-ons one by one.
- Pro: Universal-2 plus the $0.02 diarization add-on makes 1,000 file hours $170 a month
- Pro: Universal-Streaming costs the same as Universal-2, the lowest live rate in this comparison
- Pro: The free tier covers up to 185 hours of files or 333 hours of streaming, with no card
- Con: Diarization on streams costs six times the file add-on
- Con: Multichannel audio is billed per channel, so a 1-hour stereo call counts as 2 hours
This pricing model is very generous, especially the Free tier offering substantial transcription hours.
Watch out
Overage fees apply beyond free tier limits
- Starting price
- From €0.16/hr files, priced in euros
- Free plan
- Yes
- Rating
- 4.8 (23 reviews)
Best for: High-volume file transcription where an automatic volume discount beats a negotiated contract.
- Pro: Solaria-3 drops from €0.16 an hour under 100 hours to €0.12 past 10,000 hours, with no contract
- Pro: Speaker diarization, custom vocabulary and punctuation are included on files
- Pro: Top-ups of €1,000 or more earn 10% to 20% in bonus credits, and top-up credits never expire
- Con: Real-time is €0.68 an hour and has no diarization
- Con: Self-serve accounts are opted in to model training by default, with a free opt-out
Gladia's pricing is competitive, especially with its generous 10 hours free for self-serve users.
Watch out
Overage fees not explicitly stated
Azure AI Speech
A low batch rate on the Microsoft invoice, with commitment tiers for steady volume.
- 4.0 on G2 (2 reviews)
- Starting price
- $0.18/hr batch, $1/hr real-time
- Free plan
- No
Best for: Microsoft-billed teams transcribing recorded audio in batch jobs.
- Pro: Batch transcription is the lowest Microsoft rate on the Standard pay-as-you-go table, with no commitment
- Pro: The free tier includes 5 real-time audio hours a month
- Pro: A commitment of $1,600 a month covers 2,000 hours, with overage at $0.80 an hour
- Con: Real-time transcription is $1 an hour, more than five times the batch rate
- Con: Diarization and language identification on real-time cost $0.30 an hour per feature
Azure Speech's pay-as-you-go model is fair and competitive, aligning with industry standards for AI services.
Google Cloud Speech-to-Text
The cheapest Google rate comes from dynamic batch, which trades processing urgency for a lower price.
- 4.6 on Capterra (556 reviews)
- Starting price
- $0.003/min dynamic batch
- Free plan
- No
- Rating
- 4.6 (556 reviews)
Best for: Google Cloud customers with file backlogs that can wait for low-urgency processing.
- Pro: Dynamic batch on the V2 API is $0.003 a minute, under a fifth of the standard rate
- Pro: Standard recognition falls from $0.016 to $0.004 a minute past 2 million monthly minutes
- Pro: Billing is per second, and new Google Cloud accounts get $300 in free credits
- Con: Standard recognition, the rate for live audio, is $0.016 a minute, $0.96 an hour
- Con: Each audio channel is billed separately, so 30 seconds of 4-channel audio bills as 120 seconds
Google Cloud is the third-largest cloud provider (12% market share behind AWS at 31% and Azure at 25%) but leads in two areas: data analytics (BigQuery is best-in-class) and serverless containers (Cloud Run).
OpenAI transcription API
Token-billed transcription models, each with a printed per-minute estimate on the pricing page.
- 4.3 on G2 (52 reviews)
- Starting price
- From $0.003/min (estimated)
- Free plan
- Free to use
- Rating
- 4.3 (52 reviews)
Best for: Teams already on the OpenAI API who want transcription on the same key and bill.
- Pro: gpt-4o-mini-transcribe is the cheapest of the four transcription models on the page
- Pro: gpt-transcribe is estimated at $0.0045 a minute, with gpt-4o-transcribe a step above it
- Pro: Live transcription and live translation models sit on the same pricing page at $0.017 and $0.034 a minute
- Con: Billing runs on tokens ($1.25 input and $5 output per 1M for the mini model), so the per-minute figure is an estimate
- Con: Live transcription at $0.017 a minute, $1.02 an hour, is the highest live rate in this table
Whisper's pricing is incredibly generous, especially with its robust open-source option.
- Starting price
- $0.20/hr Reverb English
- Free plan
- No
- Rating
- 4.7 (563 reviews)
Best for: English-first workloads that may need an occasional human transcript at a known per-minute price.
- Pro: Reverb is $0.20 an hour in English and $0.30 in 57 other languages
- Pro: Human transcription is listed at $1.99 a minute on the same page
- Pro: New accounts get credits worth 5 hours of Reverb, usable on every product
- Con: Every request is billed at least 15 seconds, which adds up on short voice clips
- Con: Whisper Large on Rev AI is $0.005 a minute, above the Reverb rate
Rev AI offers a highly granular and competitive pricing structure, especially for its AI-powered transcription services.
Watch out
Minimums: 15 seconds for AI, 1 minute for human.
ElevenLabs Scribe
Scribe v2 for files and Scribe v2 Realtime for live audio, with a small gap between the two rates.
- Starting price
- $0.22/hr files, $0.39/hr live
- Free plan
- Yes
- Rating
- 4.6 (1,236 reviews)
Best for: Voice products already using ElevenLabs speech synthesis that want one vendor for both directions.
- Pro: Scribe v2 Realtime at $0.39 an hour is under half of Amazon's or Google's live rate
- Pro: The Pro plan at $99 a month includes 450 hours of Scribe v2
- Pro: Scribe v2 Medical costs the same $0.22 an hour and is HIPAA-eligible with BAAs
- Con: Keyterm prompting adds $0.05 an hour on files and $0.08 on live audio
- Con: Transcript editing is another $0.07 an hour on files, and more on live audio
The main gotcha: credit consumption varies by model, Flash/Turbo models cost 0.5 credits per character, but the flagship Multilingual v2 costs 1:1, so your effective minutes depend heavily on which voice model you use.
Watch out
Credit consumption varies by model: Multilingual v2 uses 1 credit per character, but Flash/Turbo models use only 0.5 credits per character. Choosing the right model can double your effective quota.
Speechmatics
Diarization and most core features included, with list rates that halve if you opt into training and commit to volume.
- 4.7 on G2 (70 reviews)
- 3.7 on Trustpilot (1 reviews)
- Starting price
- $0.24/hr list, Melia 1 files
- Free plan
- Yes
- Rating
- 4.6 (71 reviews)
Best for: Multilingual or mixed-language audio where included diarization and code-switching matter more than the base rate.
- Pro: Speaker diarization, smart formatting and word timings are included in the hourly rate
- Pro: The headline rate is the $0.24 list rate after a 33% training discount and a 25% volume discount (0.24 x 0.67 x 0.75 = 0.12, our arithmetic)
- Pro: Melia 1 transcribes 56 languages and switches language mid-sentence
- Con: The model-training discount means letting Speechmatics use your audio, and it is off by default
- Con: Real-time Enhanced is $0.80 an hour at list, and SSO is a $280 a month add-on
Speechmatics offers a generous Free tier with 480 minutes of Speech-to-Text and 1 million characters of Text-to-Speech, making it very accessible for developers and small projects.
Watch out
Specific language support for TTS is limited to English in Free/Pro.
Amazon Transcribe
Per-second billing with diarization and custom vocabulary included, at the highest batch rate in this table.
- 4.4 on G2 (17,730 reviews)
- Starting price
- $0.006/min batch, US East
- Free plan
- No
- Rating
- 4.4 (17,730 reviews)
Best for: AWS-native teams that want transcription inside their account, IAM and invoice.
- Pro: Speaker diarization, custom vocabularies and language identification are in the base rate
- Pro: Up to two channels per file or stream are billed as one, so a stereo call is not doubled
- Pro: Billing is per second with no minimum, and the free tier covers 60 minutes a month for 12 months
- Con: Batch at $0.006 a minute is $0.36 an hour, double AssemblyAI's Universal-2
- Con: PII redaction adds $0.0024 a minute on the first 250,000 minutes
AWS dominates cloud infrastructure with 31% market share and the broadest service catalog (200+ services), but that breadth is also its biggest cost trap.
Watch out
Free tier expiration trap: EC2 (750hrs t3.micro), RDS (750hrs db.t3.micro), and S3 (5GB) free tiers expire after 12 months. Many startups budget assuming free tier and get surprised by a 3-10x cost jump at month 13. Lambda and DynamoDB free tiers are permanent.
- Starting price
- Check the current rate on Deepgram
- Free plan
- Yes
- Rating
- 4.6 (475 reviews)
Best for: Voice-agent builders who need turn detection and will compare a quote against the rates above.
- Pro: Pay As You Go starts with a $200 credit, no credit card, no minimums and no expiration
- Pro: Growth plans prepay credits from $4K a year and save up to 20%
- Pro: Nova-3 covers 50+ languages with diarization and keyterm prompting; Flux adds end-of-turn detection
- Con: Per-minute rates are not in this comparison; check the current price on Deepgram's site
- Con: Flux runs only on the v2 listen endpoint, so moving from Nova-3 means a code change
Deepgram is the best-value speech AI platform for developers, and it is not close.
Watch out
Whisper Cloud (self-hosted Whisper model via Deepgram) is limited to 5 concurrent connections on all plans. If you need Whisper specifically at scale, you may need to self-host.
Other Transcription tools worth considering
More published tools from our Transcription category, ordered by our category ranking. They are not part of the editorial picks above.
How speech-to-text APIs are priced
A speech-to-text API takes audio and returns text, either from a finished file (batch, also called pre-recorded or async) or from a live stream over a WebSocket (streaming or real-time). Vendors bill by the length of the audio, usually per second, and print the rate per hour or per minute.
Four things change the rate. The mode: live audio holds a connection open, so it costs more than a file. The model tier: AssemblyAI sells Universal-3.5 Pro at $0.21 an hour, above its Universal-2 rate. Add-ons: speaker diarization, PII redaction, keyterm prompting and summaries are priced per hour on top at several vendors. Channels: Google Cloud bills each audio channel separately, and AssemblyAI bills a 1-hour stereo file as 2 hours, while Amazon Transcribe includes up to two channels in the price.
OpenAI is the exception to per-second billing. Its transcription models are billed on tokens, and the pricing page prints an "estimated cost" per minute next to the token rates.
Why the cheapest rate depends on the job
Below AssemblyAI, the cheapest file APIs are not the cheapest live ones, so a product that does both should price each mode separately. Azure's real-time rate is $1 an hour, more than five times its batch rate, while ElevenLabs keeps the gap small at $0.22 for files and $0.39 live.
Speaker labels are the add-on that most often changes a ranking. AssemblyAI charges $0.02 an hour for diarization on files and $0.12 on streams, and Azure prices real-time diarization as a separate per-feature add-on. Speechmatics, Gladia (files only) and Amazon Transcribe include it in the base rate, so a call-center workload that always needs labels compares differently from a podcast archive that never does.
Volume is the third lever. Google's standard recognition drops from $0.016 a minute to $0.004 past 2 million minutes a month, and Gladia's file rate falls step by step from €0.16 to €0.12 an hour as monthly volume grows. At 50 hours a month none of that matters; at 20,000 it decides the bill.
Key Features to Look For
Batch rate per audio hour (Essential)
The list price for recorded files on the cheapest general model, the number most bills are built on.
Streaming rate per audio hour (Essential)
The price for live audio over a WebSocket, which costs more than the file rate everywhere here except AssemblyAI.
Speaker diarization cost (Essential)
Included at Speechmatics and Amazon Transcribe, a per-hour add-on at AssemblyAI and on Azure real-time.
Billing increment and minimum (Important)
Per second with no minimum at Amazon Transcribe and Google; Rev AI rounds up to the second with a 15-second minimum per request.
Channel billing (Important)
Whether a stereo call is billed as one hour or two. AssemblyAI and Google bill per channel.
Free allowance (Nice to have)
Signup credits or monthly free minutes, useful for a pilot but excluded from the bills on this page.
What to Consider
-
Decide between files and live audio first: Azure's live rate is more than five times its file rate, so the mode matters more than the vendor
-
Price speaker labels into the rate if your audio has more than one voice, since AssemblyAI's streaming diarization add-on costs six times its file add-on
-
Check how stereo is billed: AssemblyAI bills a 1-hour, 2-channel file as 2 hours, while Amazon Transcribe includes up to two channels
-
For short clips, read the minimum charge: Rev AI bills at least 15 seconds per request
-
Gladia bills in euros, so a US budget carries an exchange-rate line
-
Data use is a price lever at Speechmatics (33% off when you let it train on your audio) and a default at Gladia self-serve (training on, free opt-out)
Evaluation Checklist
Run the same 30 minutes of your own audio through two shortlisted APIs on their free credits and compare word errors on names and numbers
Confirm the streaming rate for the exact model you will deploy, not the batch rate on the same card
List every add-on your product needs (diarization, redaction, keyterms) and add each per-hour price to the base rate
Work out your monthly hours and check which volume tier applies at Google or Gladia
Ask whether multichannel audio is billed per channel
Pricing Overview
Recorded files, under $0.25 an hour
Podcasts, meeting archives and support-call backlogs where a transcript can arrive minutes later. AssemblyAI, Azure, Google dynamic batch, OpenAI, Rev AI, ElevenLabs and Speechmatics.
$0.15 to $0.24 per audio hour
Live audio, under $0.50 an hour
Voice agents and live captions on AssemblyAI Universal-Streaming, ElevenLabs Scribe v2 Realtime or Speechmatics Melia 1 real-time.
Up to $0.45 per audio hour
Live audio on the big clouds
Teams already billed by AWS, Google Cloud, Azure or OpenAI who want transcription on the same invoice and accept a higher live rate.
$0.60 to $1.02 per audio hour
Pricing Comparison
| API | Batch model and rate | Batch per hour | Streaming per hour | Speaker diarization | 1,000 batch hours a month | Free allowance |
|---|---|---|---|---|---|---|
Universal-2 | $0.15 ($0.21 on Universal-3.5 Pro) | Same as batch on Universal-Streaming; $0.45 Universal-3.6 Pro Realtime | +$0.02/hr on files, +$0.12/hr on streams | $170 with diarization | Up to 185 hours of files or 333 hours of streaming | |
Gladia | Solaria-3, priced in euros | €0.16 under 100 hours, down to €0.12 past 10,000 | €0.68 | Included on files; files only | €140 (500 to 2,000-hour tier) | €50 in credits, valid one year |
Standard batch transcription | $0.18 | $1.00 | Real-time add-on $0.30/hr per feature | $180 | 5 real-time hours a month on the free tier; batch not included | |
V2 dynamic batch, $0.003/min | $0.18 (standard recognition $0.96) | $0.96 (standard rate, first 500,000 minutes) | Not priced separately | $180 dynamic batch; $960 standard | V1: 60 minutes a month; $300 Google Cloud credits | |
OpenAI | gpt-4o-mini-transcribe, $0.003/min estimated | $0.18 (gpt-transcribe $0.27) | $1.02 (gpt-live-transcribe) | Not priced separately | $180, estimated from token pricing | None listed on the pricing page |
Reverb, English | $0.20 ($0.30 other languages) | Check Rev AI's site | Not priced separately | $200 | 5 hours of Reverb in credits | |
ElevenLabs | Scribe v2 | $0.22 | $0.39 (Scribe v2 Realtime) | Not priced separately | $220 | Free plan includes about 4.5 hours |
Speechmatics | Melia 1, pay as you go | $0.24 list; $0.12 with both discounts | $0.40 list | Included | $240 at list | $100 in credits |
Standard batch, US East | $0.36 ($0.006/min) | $0.60 ($0.01/min) | Included | $360 | 60 minutes a month for 12 months | |
Deepgram | Nova-3 | Check Deepgram's site | Check Deepgram's site | Check Deepgram's site | Check Deepgram's site | $200 in credits, no card |
List prices checked on October 28, 2026 on each vendor's own page: AssemblyAI pricing, Gladia pricing, Azure Speech pricing, Google Cloud Speech-to-Text pricing, OpenAI API pricing, Rev AI pricing, ElevenLabs API pricing, Speechmatics pricing, Amazon Transcribe pricing and Deepgram's plan summary. Hourly figures for vendors that print per-minute rates are our arithmetic (rate times 60). Bills exclude free credits and taxes.
Mistakes to Avoid
- ×
Comparing a vendor's batch rate with another vendor's streaming rate: Azure's two rates differ more than fivefold
- ×
Forgetting that stereo audio can bill twice: AssemblyAI and Google count each channel
- ×
Reading Speechmatics' headline rate as the pay-as-you-go price, when the list rate for Melia 1 files is twice that
- ×
Treating OpenAI's per-minute figures as fixed prices, when billing runs on tokens and the page calls them estimates
- ×
Budgeting from free credits: AssemblyAI's 185 free hours and Gladia's €50 cover a pilot, not a month in production
Expert Tips
- →
Send recorded audio through batch even inside a live product when the user can wait: on Azure that cuts the rate by more than 80%
- →
If you need speaker labels on live calls, compare AssemblyAI at $0.27 an hour all-in (Universal-Streaming plus its diarization add-on) with Speechmatics at $0.40, where labels are included
- →
Split stereo call recordings yourself only when the vendor bills per channel and you need per-speaker accuracy; otherwise send mono
- →
On Google, enable dynamic batch for any backlog: the same audio costs under a fifth of the standard rate
- →
Re-price at each Gladia tier boundary (100, 500, 2,000, 5,000 and 10,000 hours a month), since the whole month moves to the new rate
Red Flags to Watch For
-
A headline rate that only applies after an opt-in to model training or a volume commitment
-
Per-minute estimates on token-billed models with no cap on output tokens
-
A 15-second or longer minimum charge on a product that sends many short clips
-
Diarization priced separately on streams but included on files, so the live bill grows faster than expected
The Bottom Line
Start with AssemblyAI if you want the lowest dollar rate for both recorded files and live audio, one hourly rate for each, with every add-on priced on the page. Pick Gladia when volume is high and euro billing is acceptable, since its rate falls on its own as hours grow. Stay on Azure, Google or OpenAI for files when one cloud invoice matters more than a few cents an hour, but move live audio elsewhere, because their live rates run $0.96 to $1.02 an hour. Choose Speechmatics or Amazon Transcribe when included diarization outweighs the higher base rate, and get a current Deepgram quote before deciding on voice agents.
Frequently Asked Questions
How much does a speech-to-text API cost per minute in 2026?
List rates for recorded audio run from $0.0025 a minute (AssemblyAI Universal-2) to $0.006 a minute (Amazon Transcribe), as of October 28, 2026. Live audio runs from $0.0025 a minute on AssemblyAI Universal-Streaming to $0.017 a minute on OpenAI's gpt-live-transcribe.
What does 1,000 hours of transcription cost through an API?
On recorded audio at list prices, 1,000 hours a month costs $170 on AssemblyAI with speaker labels, €140 on Gladia, $180 on Azure or Google dynamic batch, $200 on Rev AI, $220 on ElevenLabs, $240 on Speechmatics and $360 on Amazon Transcribe (hours times each hourly rate, our arithmetic).
Is streaming transcription more expensive than batch?
Yes at almost every vendor. Azure charges $1 an hour for real-time, more than five times its batch rate, and Amazon Transcribe charges two-thirds more for live audio than for files. AssemblyAI is the exception at the low end, with Universal-2 and Universal-Streaming at the same price.
Does speaker diarization cost extra?
It depends on the vendor. AssemblyAI adds $0.02 an hour on files and $0.12 on streams, and Azure adds $0.30 an hour per feature on real-time. Speechmatics and Amazon Transcribe include diarization in the base rate, and Gladia includes it on files only.
Is there a free speech-to-text API?
Several APIs start free but none stays free at volume (for apps with a free plan, see the best free transcription tools). AssemblyAI's free tier covers up to 185 hours of files, Deepgram gives a $200 credit, Speechmatics $100, Gladia €50 for a year, Azure 5 real-time hours a month and Amazon Transcribe 60 minutes a month for 12 months.
Why does Speechmatics advertise $0.12 an hour?
That figure applies two optional discounts to the $0.24 pay-as-you-go rate for Melia 1 files: 33% off for letting Speechmatics train on your audio and up to 25% off for volume commitments. Without either, a pay-as-you-go account pays the full list rate.
AssemblyAI vs Deepgram: which costs less?
AssemblyAI lists the same hourly rate for Universal-2 files and Universal-Streaming as of October 28, 2026, the lowest in this comparison. Deepgram's per-minute rates are not in this comparison, so check the current price on Deepgram's site; its Pay As You Go plan starts with a free credit. See the AssemblyAI alternatives for more options.
Cite this page: Toolradar, "Speech-to-Text API Pricing Compared in 2026", updated October 2026, https://toolradar.com/guides/speech-to-text-api-pricing-compared
Related Guides
Ready to Choose?
Compare features, read reviews, and find the right tool.
Some offers on this page may be paid placements or contain affiliate links.