Fireflies.ai Gives Away Transcription and Meters the Storage
There is a neat piece of positioning at the top of Fireflies’ pricing page: unlimited transcription and unlimited AI summaries, on every plan including the free one. It is true, and it is also not the whole picture. What Fireflies meters is storage, counted in minutes, and AI credits, counted in units the vendor does not define in dollar terms. A free workspace gets 400 minutes of storage for the entire team, which is roughly seven hours of retained meetings before something has to go.
Fireflies.ai plan facts, USD, checked August 2026
- Free provides 400 storage minutes per team and 20 AI credits
- Pro is $18 per seat monthly or $10 annually, with 8,000 storage minutes per seat
- Business is $29 per seat monthly or $19 annually, with unlimited storage
- Enterprise is $39 per seat and is sold on annual billing only
- Annual commitment is advertised at up to 44% below monthly rates
Unlimited transcription, 400 minutes of memory
The distinction between transcribing and retaining is the one that matters. Fireflies will happily process every call you throw at it on the free plan, but the archive it keeps is capped at 400 minutes across the whole team, not per person. Two people with a normal meeting load will fill that in under a week.
Pro raises storage to 8,000 minutes per seat, which is around 133 hours and is comfortably beyond what most individuals generate. Business removes the cap entirely. Framed properly, the tiers are not really about transcription quality at all, they are about how long you are allowed to remember things.
There is a second reason the storage figure matters more than it first appears. Fireflies is most valuable when it functions as a searchable archive, letting someone answer what a client said about pricing four months ago. A team on the free plan with 400 shared minutes never accumulates enough history for that use case to exist, which means the free tier demonstrates transcription while withholding the capability that actually justifies the product.
Twenty, thirty, fifty: the credit ladder
AI credits are allocated at 20 on Free and Pro, 30 on Business and 50 on Enterprise. Fireflies uses them for the generative work that sits on top of the transcript, the ask-anything queries, the custom summary formats and similar.
The published material does not translate a credit into a specific number of operations, which makes the ladder difficult to evaluate before purchase. That Pro and Free receive the same 20 credits is itself informative: the paid upgrade from Free is about storage and recording features, not about additional AI headroom, and anyone upgrading specifically for more generative capacity is buying the wrong tier.
What Business adds beyond the storage cap
- Multi-language mode, which matters for teams running calls in more than one language
- Conversation intelligence, including talk-time ratios and topic tracking across meetings
- Team analytics that aggregate across seats rather than reporting per individual
- Public access controls for sharing meeting records outside the workspace
- Unlimited retention, removing the storage-minute calculation entirely
Accents, crosstalk and the speaker-label problem
Comparative coverage of Fireflies against Otter tends to land in the same place: both perform respectably on clear audio and both degrade on the same inputs. Strong regional accents, rapid crosstalk and specialist vocabulary are where the word error rate climbs, and no vendor in this category publishes accuracy figures broken down by accent, which is itself a meaningful omission.
Speaker diarisation remains the weakest link across the whole product class. When Fireflies mislabels who said what, the error propagates into the summary and the extracted action items without any confidence signal. A summary that confidently assigns a commitment to the wrong person is worse than no summary, and it is the failure mode that most often ends a trial.
The consent question every notetaker now inherits
The privacy litigation consolidated against Otter.ai in the Northern District of California during 2025 was not framed by commentators as a problem unique to one vendor. Employment lawyers writing about the case treated it as a warning to any organisation deploying an automated notetaker that joins meetings from a calendar sync.
Fireflies operates the same basic pattern, and the same mitigations apply: announce recording verbally, disable automatic joining for externally-facing staff, and check your obligations in two-party consent jurisdictions. None of this is a criticism of Fireflies specifically. It is a criticism of buying any tool in this category without a recording policy written first.
Enterprise at $39 and the annual-only condition
Enterprise carries the controls that make the tool deployable in a regulated environment: SSO with SCIM provisioning, audit logs, HIPAA compliance and private storage, plus a named account manager. At $39 per seat it is not unreasonably priced against comparable compliance-grade tooling.
The condition is that it is offered on annual billing only, so there is no low-commitment route to evaluating it. Any organisation whose procurement depends on the HIPAA or audit features cannot pilot them monthly, which pushes the evaluation burden onto a sales process rather than a trial. Ask for a written commitment on data residency at the same time, because private storage and geographic residency are not the same guarantee and the pricing page does not distinguish them.
One further point applies across every tier. Fireflies’ quality depends heavily on the audio it receives, and it receives whatever the meeting platform sends. A participant on a poor connection, a room microphone picking up three people at a distance, or a call where several attendees dial in by phone will produce a transcript materially worse than the vendor’s demonstrations, regardless of what you are paying. This is not a fault in the product so much as a limit on what any tool in the category can promise, and buyers should test on their own worst-case call rather than a clean one.
Sources: Fireflies.ai pricing · G2, Fireflies.ai compared with Otter.ai · National Law Review, privacy litigation and AI notetakers · Sybill, Fireflies against Otter.ai comparison
