Pricing
The two price the same call differently, so headline figures are not directly comparable: Retell quotes one range that spans its model and voice choices, and Dialgem adds four parts. Use each side's calculator with identical choices.
| Retell AI (published 31 Oct 2026) | Dialgem | |
|---|---|---|
| Headline | $0.07 to $0.31 per minute for AI voice agents | $0.045 to $0.135 per connected minute, all four parts |
| Platform | Voice infrastructure $0.055 per minute | $0.030 per minute (speech recognition, recording, post-call analysis) |
| Voice | $0.015 per minute (platform voices, Cartesia, OpenAI and others) up to $0.10 (ElevenLabs v3 and v4) | $0.010 to $0.035 per minute (self-hosted, Cartesia, ElevenLabs) |
| Language model | From $0.0016 to $0.32 per minute depending on model | $0.005 to $0.035 per minute |
| Phone line | $0.015 per minute on Retell numbers; no charge for your own SIP trunk | $0.025 inbound, $0.035 outbound; $0 on your own SIP trunk or web calls |
| Add-ons | Per-minute options such as knowledge base (+$0.005), PII removal (+$0.01), denoising (+$0.005); batch calls +$0.005 per dial | The calculator lists the four parts per minute |
| Billing | Calls tracked to the nearest second | Billed per second from prepaid credit |
Features
Where a row says both, either platform does it. Rows are limited to things we have checked on our side.
| Retell AI | Dialgem | |
|---|---|---|
| Interruption handling | Described as turn-taking and interruptions handled like a person | Stops mid-sentence when the caller speaks; sensitivity and responsiveness are per-agent settings |
| Voicemail | Detects voicemail in the first 3 minutes of a call; hang up or leave a message | Answering-machine detection from audio plus a transcript hint; hang up or leave a message |
| Agent types | Single prompt or conversation flow | Prompt agents and conversation-flow agents |
| Knowledge base | Crawl a website or upload documents | A knowledge base per agent |
| Outbound at scale | Batch calls, immediate or scheduled | Campaigns |
| Your own SIP trunk | Yes; elastic SIP trunking is the option it recommends | Yes |
| SDKs | Official Node.js and Python libraries | TypeScript SDK; other languages call the REST API |
| Language models | A long list including GPT, Claude and Gemini families and speech-to-speech | Anthropic or OpenAI, with a configurable fallback |
| Voices | Retell platform voices, Cartesia, ElevenLabs, OpenAI, Minimax, Fish, Inworld | Cartesia, ElevenLabs or self-hosted |
| Compliance | States HIPAA with a BAA, SOC 2 Type 1 and 2, and GDPR | No SOC 2 report or HIPAA attestation today; no BAA |
How to read this
If you need HIPAA, a Python SDK or the widest choice of models and voices, Retell is ahead on every one. If you want a rate card you can read in one glance, prepaid credit without a contract, and an API whose shapes match what a Retell integration already uses, Dialgem is worth a test call. The rest is close enough that your own calls will tell you more than this table.
Questions
- How current is this?
- Retell's figures were read from its public pages on 31 October 2026. Vendors change prices and features, so check the sources linked below before relying on any number.
- Why are call transfers and keypad handling not in the table?
- Because we only list what we have verified on our side. Both platforms describe them; compare them yourself on a test call.
Sources
This is an independent comparison written by Dialgem. Dialgem is not affiliated with or endorsed by Retell AI; Retell is a trademark of its owner and is used here only to identify its product. Retell's details were read from its public pages on 31 October 2026 and can change. Check the sources before you decide, and tell us if anything is out of date.