Unlimited vs Metered Transcription Pricing: How to Pick in 2026

Flat-rate unlimited wins once you exceed roughly 20 hours of audio per month. Below that, metered per-minute APIs are cheaper. This post shows the exact numbers so you can pick without guessing.

The Two Models Defined

Flat-rate unlimited means a fixed monthly fee with no per-minute charge. Your unit cost drops as usage rises because the fee stays constant. Examples in 2026:

• CATT Pro: $9.99/month (annual) or $14.99/month (monthly), unlimited transcription.
• TurboScribe Unlimited: $10/month (annual) or $20/month (monthly), unlimited transcription.
• Otter.ai Business: $19.99/user/month (annual), unlimited meeting recordings (6,000 min/user cap on imported files).

Metered per-minute means you pay only for what you process. No monthly minimum, no churn cost. Examples with 2026 verified rates:

• Deepgram Nova-3: $0.0077/min ($0.46/hr) pay-as-you-go.
• AssemblyAI Universal-2: $0.0025/min ($0.15/hr).
• OpenAI Whisper-1 / gpt-4o-transcribe: $0.006/min ($0.36/hr).
• Rev.ai Reverb: $0.0033/min ($0.20/hr).

Note: metered rates above are base transcription only. Speaker diarization, sentiment analysis, and topic detection stack on top as add-ons. For a deeper look at API rate structures, see transcription pricing models explained.

The Break-Even Math

The break-even is the monthly volume at which both models cost the same. Above it, flat-rate wins. Below it, metered wins.

CATT Pro ($9.99/month) vs Deepgram Nova-3 ($0.0077/min):

Break-even = $9.99 / $0.0077 = 1,297 minutes = 21.6 hours per month.

CATT Pro vs AssemblyAI Universal-2 ($0.0025/min):

Break-even = $9.99 / $0.0025 = 3,996 minutes = 66.6 hours per month.

AssemblyAI's base rate is low enough that metered beats flat-rate for most individual users. The catch: diarization adds $0.02/hr, sentiment adds $0.02/hr, and so on. A fully-featured job matches closer to $0.20-0.25/hr, which brings the break-even down to roughly 40-50 hours.

CATT Pro vs Rev.ai Reverb ($0.0033/min):

Break-even = $9.99 / $0.0033 = 3,027 minutes = 50.4 hours per month.

At these rates, CATT is cheaper for anyone above 50 hours/month, while Rev.ai undercuts it below that. Rev.ai's rates have dropped considerably from the per-minute model it used a few years ago.

Monthly Cost at 2, 10, and 40 Hours

The chart below compares CATT Pro (flat $9.99/month, annual billing) against Deepgram Nova-3 ($0.0077/min) at three usage levels. At 40 hours per month, the gap is clear.

Deepgram @ 2h

$0.92

CATT @ 2h

$9.99

Deepgram @ 10h

$4.62

CATT @ 10h

$9.99

Deepgram @ 40h

$18.48

CATT @ 40h

$9.99

CATT Pro @ $9.99/mo (annual) vs Deepgram Nova-3 @ $0.0077/min. Verified July 2026.

At 2 hours per month, Deepgram costs $0.92 versus $9.99 flat-rate. At 10 hours, $4.62 versus $9.99. At 40 hours, the metered bill is $18.48 and the flat-rate bill is still $9.99. The crossover for this specific pair is 21.6 hours.

When Metered Pricing Is the Right Call

Three situations where per-minute billing genuinely wins.

Very low volume. If you transcribe under 10 hours per month, you are unlikely to break even on most flat-rate plans. Paying $0.92 per month on Deepgram beats paying $9.99 flat-rate for light use. The CATT free tier (10 minutes per month) covers the lightest occasional needs without any cost at all.

Highly variable volume. If your usage swings from zero one month to 80 hours the next, metered pricing lets you pay nothing in the quiet months. Annual flat-rate subscriptions (often 30-50% off monthly rates) partially offset this, but you still pay for every month you subscribed.

Building a product. If transcription is a feature inside your own software, metered pricing aligns your cost to your revenue per job. A flat-rate plan would mean absorbing variable costs or building a custom usage layer on top. For builders, see the comparison in speech-to-text API pricing 2026.

When Flat-Rate Unlimited Is the Right Call

Most consumer and prosumer workflows.

Consistent moderate-to-high volume. Any user consistently above their specific break-even point saves money every month. The savings compound significantly at higher volumes: at 100 hours per month, CATT at $9.99 implies a $0.00167/min effective rate, well below any API's list price.

Predictable budget. A known monthly cost makes financial planning simpler. This matters more for teams with expense approval processes than for solo users, but it is a real advantage.

Feature bundling. Consumer flat-rate plans typically include a UI, editor, speaker labels, AI summaries, and export formats in the subscription. Building equivalent functionality on top of a raw API adds real engineering cost that the math above ignores. If you would otherwise pay for those features separately, the effective break-even on flat-rate shifts lower.

Experimentation. Re-running the same audio for different export formats or correcting speaker labels does not cost extra on a flat-rate plan. On metered pricing, every re-process is a new charge.

The Hidden Variables in the Comparison

A few factors the simple math misses.

Add-ons stack. Metered API headlines are base transcription only. On AssemblyAI, diarization adds $0.02/hr. On Deepgram, diarization and keyterm prompting are separate line items. On a feature-comparable basis, the effective metered rate can be 30-100% above the base figure. Consumer flat-rate plans bundle most of these features into the subscription price. The hidden costs of transcription services post has a full breakdown.

Fair-use clauses. Unlimited flat-rate plans include fair-use terms. For typical individual or small-team workloads (5-150 hours per month), this rarely matters. At agency or network scale, it is worth reading the terms before committing.

Volume discounts on metered. Deepgram's Growth plan drops Nova-3 to $0.0065/min from $0.0077/min, but requires a $4,000+ annual commitment. At enterprise volume with negotiated rates, metered pricing can beat flat-rate again. See enterprise transcription pricing for that case.

Who Should Choose Which

My take: for anyone who transcribes more than three to four hours of audio per week with any regularity, a flat-rate consumer plan almost always wins on cost. The uncertainty shows up only at the edges: very low volume, very irregular use, or pure API-integration use cases.

A Worked Annual View

Annual numbers make the trade-off concrete. Assume CATT Pro at $9.99/month annual ($119.88/year) versus Deepgram Nova-3 at $0.0077/min:

• 5 hours/month: Deepgram = $27.72/year. Flat-rate = $119.88/year. Metered saves $92.
• 20 hours/month: Deepgram = $110.88/year. Flat-rate = $119.88/year. Metered saves $9 (essentially a tie).
• 40 hours/month: Deepgram = $221.76/year. Flat-rate = $119.88/year. Flat-rate saves $102.
• 100 hours/month: Deepgram = $554.40/year. Flat-rate = $119.88/year. Flat-rate saves $435.

The cross-over sits at roughly 21-22 hours per month for this pair. Readers who are unsure of their volume should track it for one month. The CATT free tier (10 minutes per month, no card required) covers light use during that measurement period. See also free vs paid transcription services for the broader landscape.

If you just need a clean transcript without meeting-bot infrastructure, a monthly limit, or per-job billing, CATT's audio-to-text tool works on upload or URL, and the $9.99/month annual plan removes all caps.

Frequently Asked Questions

What is the break-even point between unlimited and metered transcription?

It depends on the specific plans you compare. Against Deepgram Nova-3 at $0.0077/min, the break-even with a $9.99/month flat-rate plan is roughly 1,297 minutes (about 21.6 hours per month). Against AssemblyAI Universal-2 at $0.0025/min, the break-even rises to 3,996 minutes (about 66.6 hours per month) because the metered rate is so low. Always run the math against the specific services you are evaluating.

Does "unlimited" actually mean no limits?

Usually not in an absolute sense. Most unlimited flat-rate plans include a fair-use clause in their terms of service. TurboScribe and CATT both state no hard per-month cap, but heavy commercial use (think podcast networks processing thousands of hours) can trigger a conversation with support. For typical individual or small-team workloads, the plans genuinely work as advertised.

When does metered per-minute pricing make more sense?

Three situations: (1) you transcribe fewer than 10-20 hours per month and do not need a consumer UI, (2) your volume is highly unpredictable with several zero-usage months per year, or (3) you are building a product and need per-minute cost to align with your unit economics. In all three cases, a developer API with no monthly minimum is structurally the better fit.

Are there hidden costs in metered transcription APIs?

Yes. Speaker diarization, sentiment analysis, topic detection, and translation are usually priced as add-ons on top of the base transcription rate. AssemblyAI, for example, adds $0.02/hr for standard diarization on top of its $0.15/hr base rate. Deepgram lists diarization and keyterm prompting as separate line items. These stacked costs can push the effective rate 30-100% above the headline number. Consumer flat-rate plans typically bundle diarization and exports into the subscription price.

Can I start metered and switch to flat-rate later?

Yes, and this is actually a sensible approach. Use a metered service or the free tier of a consumer tool for a month or two to measure your real volume. Once you have data, the math becomes straightforward. Most flat-rate plans bill monthly with no long-term lock-in (annual plans save 30-50% but require commitment), so switching costs are low if your usage pattern changes.

Sources

• Deepgram Pricing - Nova-3 pay-as-you-go and Growth plan rates verified July 2026.
• TurboScribe Pricing - Monthly and annual Unlimited plan rates.
• AssemblyAI Pricing - Universal-2 base rate and add-on pricing.
• Rev.ai Pricing - Reverb Transcription hourly rate.
• Otter.ai Pricing - Pro and Business plan tier structure and limits.
• OpenAI Transcription Pricing - Whisper-1 and gpt-4o-transcribe per-minute rates.
• Deepgram Pricing 2026 (diyai.io) - Nova-3 rate cross-reference.
• BrassTranscripts Deepgram Pricing - Pay-as-you-go vs Growth plan comparison.