✓ Last checked: 2026-08-31· Source: official vendor pricing pages· report a change →
Building an AI meeting notetaker, call-recording feature or transcript pipeline on top of Zoom/Meet/Teams needs a meeting bot API to actually join the call. Recall.ai charges a flat $0.50 per bot-hour with no subscription, Meeting BaaS sells hour-tokens ($0.35–$0.50) plus optional monthly plans for higher bot-concurrency, and Vexa.ai undercuts both at $0.30/hr pay-as-you-go — or $0 in vendor fees if you self-host its open-source stack. Model your own bot-hours, transcription need and storage to see which one actually wins.
Cheapest meeting bot API as bot-hours grow
Transcription, storage and plan choices held at your current inputs — only monthly bot-hours change per row. Watch where Meeting BaaS's plan fee amortizes past Recall.ai's zero-minimum rate, and where Vexa's self-host option turns free.
Bot-hours / mo
Recall.ai
Meeting BaaS
Vexa.ai
Cheapest
Why bot-hour, token-pack and self-host pricing shapes matter
Meeting bot APIs don't compete on a single number — each vendor meters the same underlying resource (a bot occupying a call for its full duration) with a different billing shape. Recall.ai bills a flat $0.50 per bot-hour with no monthly minimum: hours × 0.50 for recording, plus hours × 0.15 if you use its built-in transcription, plus $0.05/hour for any recording retained beyond the 7-day free storage window. There's no subscription fee, which makes it simple at any volume but not the cheapest per-hour rate at scale. Meeting BaaS instead sells hour-tokens in packs from $0.50 down to $0.35 each depending on pack size, layered with optional monthly plans ($0 free / $99 Pro / $199 Scale / $299 Enterprise) that raise your daily bot-concurrency cap and retention window — real-time transcription costs an extra 0.25 tokens per hour. Vexa.ai undercuts both on its managed pay-as-you-go rate at $0.30/hour recording plus $0.20/hour for transcription (same $0.50/hr total as Recall.ai's combined rate, but split differently), and is the only one of the three that's Apache-2.0 open source — self-hosting it drops the vendor bill to $0 at any volume, at the cost of running your own bot/recording/transcription infrastructure.
That means the "cheapest" vendor flips with volume and requirements. At very low, spiky usage, Recall.ai's zero-minimum flat rate often wins since there's no plan fee to amortize. At moderate-to-high volume that needs higher daily bot-concurrency, Meeting BaaS's plan fee pays for itself and its bulk token rate ($0.35/hr) can beat both competitors. At high volume where you're willing to operate infrastructure yourself, Vexa.ai's free self-host option beats every managed rate outright. Run your own bot-hours, transcription need and self-host choice through the calculator above instead of trusting a single headline rate.
The Recall.ai vs Meeting BaaS vs Vexa Cost Calculator compares three meeting-bot-API pricing shapes at your monthly bot-hours, transcription need, storage retention and self-host choice. Recall.ai cost is (hours × $0.50) + (transcribe ? hours × $0.15 : 0) + (extra storage hours × $0.05), with no subscription minimum. Meeting BaaS cost is plan fee + (hours × token rate) + (transcribe ? hours × 0.25 × token rate : 0), where the token rate depends on the pack size you'd realistically buy and the plan fee buys a higher daily bot-concurrency cap. Vexa.ai cost is 0 if self-hosting (Apache-2.0, your own infra), otherwise (hours × $0.30) + (transcribe ? hours × $0.20 : 0) on its managed cloud. The calculator shows all three totals side by side, then a second table isolates all three vendors across a range of monthly bot-hour volumes at your current settings so you can see exactly where the cheapest option changes.
These are representative pricing shapes based on each vendor's published self-serve rates as of August 2026 (Recall.ai pay-as-you-go rate and add-ons from recall.ai/pricing; Meeting BaaS token packs and plan tiers from meetingbaas.com/pricing; Vexa.ai pay-as-you-go and self-host terms from vexa.ai/pricing), not live vendor quotes. Recall.ai's Launch and Enterprise tiers, and Meeting BaaS's Enterprise tier, use custom negotiated pricing not shown here. Self-hosting Vexa.ai eliminates vendor fees but doesn't model your own server, bandwidth or engineering-time cost — factor that in separately. Always confirm current pricing directly with the vendor before budgeting.
Frequently asked questions
What is a meeting bot API and why is it billed per hour?
A meeting bot API (Recall.ai, Meeting BaaS, Vexa.ai) sends an automated bot into a Zoom, Google Meet or Teams call to capture audio/video, transcribe speech and expose the data over a developer API — the infrastructure behind AI notetaker apps like Otter or Fireflies, but meant to be embedded in your own product. All three vendors meter it by bot-runtime hour because that's the resource they actually consume: a rendering/recording session that occupies compute and bandwidth for the full duration a bot sits in a call, regardless of how much anyone spoke. That's different from a per-minute transcription API, which only charges for audio actually processed.
Is Recall.ai cheaper than Meeting BaaS?
It depends on volume and whether you need Meeting BaaS's higher daily bot caps. Recall.ai's pay-as-you-go rate is a flat $0.50 per recorded bot-hour plus $0.15/hour for built-in transcription and $0.05/hour for storage retained beyond the 7-day free window, with no separate subscription fee. Meeting BaaS instead sells hour-tokens from $0.35 to $0.50 each depending on pack size, plus optional monthly plans ($99 Pro, $199 Scale, $299 Enterprise) that raise your daily bot-concurrency and retention limits — at large token packs its per-hour rate undercuts Recall.ai, but the monthly plan fee can tip the total the other way at low volume. Run your own bot-hours through the calculator above to see which wins at your usage.
When does Vexa.ai's self-hosted option beat the managed clouds?
Vexa.ai is Apache-2.0 licensed, so self-hosting it costs $0 in vendor fees at any bot-hour volume — you only pay for your own compute (typically a GPU or CPU box running the bot + transcription pipeline) and ops time to keep it running. That makes self-host attractive once monthly bot-hours climb high enough that Recall.ai's $0.50/hr or Vexa's own managed $0.30-0.50/hr cloud rate would exceed what a dedicated server costs — commonly in the low hundreds of bot-hours per month, though the real break-even depends entirely on your infrastructure cost and whether you have the engineering time to run it. Below that volume, or if you don't want to operate call-recording infrastructure yourself, the managed cloud rate is usually cheaper once you count engineering time.
Which meeting bot API is cheapest — Recall.ai, Meeting BaaS or Vexa.ai?
There's no fixed winner because the three price transcription and plan minimums differently. Vexa.ai's pay-as-you-go rate ($0.30/hr recording, +$0.20/hr transcription) is the cheapest managed cloud option per hour at low-to-moderate volume, and its free self-host tier beats everyone at high volume if you can run the infrastructure yourself. Recall.ai has no subscription minimum, which makes it cheapest at very low, spiky volume. Meeting BaaS wins when you need its higher daily bot-concurrency caps or its cheapest token-pack rate ($0.35/hr) at bulk volume, since its monthly plan fee amortizes across more bot-hours. Run your own bot-hours, transcription need and self-host option through the calculator above instead of assuming one vendor's headline rate applies to you.