Pricing
Start free. Upgrade when you outgrow it. One workspace per subscription — bring your team along.
Free
Try it out. Personal projects.
- 5 active recordings (delete to free a slot)
- Up to 2-minute recordings
- 10 uploads per 24 hours
- 100 frames analysed per recording
- MCP, CLI, and SDK access
- Public share links
- Community support
Pro
Solo builders, AI engineers, and power users.
- 30 hours of recording per month
- Up to 4-hour recordings
- 600 frames analysed per recording
- Everything in Free, plus:
- Email support
- Priority queue for processing
Team
Teams sharing context across humans and agents.
- 200 hours of recording per month
- Up to 8-hour recordings
- 1,200 frames analysed per recording
- Everything in Pro, plus:
- Unlimited workspace members
- SSO via Google/GitHub
- Priority email support
Need more? Email sales@iris.dev for enterprise pricing.
Prices are in USD. Taxes may apply. You can cancel any time from your billing portal.
Frequently asked
Is Iris really free to start?
Yes. The Free plan never expires and doesn't require a credit card. You can upgrade or cancel any time.
What happens when I hit my plan's limits?
On Free, you have 5 active recording slots — delete a recording to free a slot for a new one. There's also a 10-upload rolling 24-hour limit so the slot system can't be gamed. Pro and Team use a monthly minutes cap (30 hours and 200 hours respectively); new recordings pause when you hit it until the next billing cycle or you upgrade.
Why a 2-minute cap on Free?
Each recording costs us a small amount in AI vision + transcription credits. Free is for trying Iris out — Pro and Team allow recordings up to 4 and 8 hours respectively.
How is recording length measured?
By the duration of the video file you upload. Time spent processing isn't counted against your quota.
Can I self-host Iris?
Yes — Iris is open-source. Run it on your own infrastructure with the Docker image and bring your own Anthropic and Deepgram keys.
Do you train AI on my recordings?
No. Recording content is sent to Anthropic and Deepgram only to extract context for you, and is not used to train any model.