Pick Vapi if developers want to choose every model and build the agent in code, paying a $0.05-a-minute hosting fee. Pick Thoughtly if a revenue team wants leads called, texted and emailed with no code, on a plan bought through sales. Dograh, the open-source AI voice agent platform we build, lets you self-host it or run it on Dograh Cloud.
Key Takeaways
- Vapi charges $0.05 a minute for hosting, with models and phone billed on top.
- Thoughtly quotes its plans, with unlimited minutes inside a cap on concurrent calls.
- Both run only on the vendor's cloud, so neither can sit on your servers.
This comparison sits inside our guide to Thoughtly alternatives, which sets hosted platforms and open-source options against Thoughtly on hosting, models, phone numbers, price and lock-in. Here we give a verdict for each, then compare how you build agents, price, compliance and lock-in, and end with where Dograh fits.
Pick Vapi or Thoughtly
Vapi and Thoughtly solve the same job for different people, so the right pick depends on who will build and run the agent.

Pick Vapi if your developers want to choose every model and wire tools in code, and a per-minute hosting fee is fine. You can bring your own models for any part of the call, and if you handle patient data you can add HIPAA, the Health Insurance Portability and Accountability Act, for $2,000 a month.
Pick Thoughtly if a revenue team wants every inbound lead called, texted and emailed, with results written back to the CRM and nobody writing code. Its model choices are GPT, Claude or Gemini, with your own model reserved for Enterprise. You buy one of its three plans through its sales team.
If you want either style on servers you control, neither fits, and the last section covers that case.
Vapi vs Thoughtly in one table
Dograh sits in the last column as the open-source option. Latency is left out, because it depends on your models and phone path more than on the platform.
| Vapi | Thoughtly | Dograh | |
|---|---|---|---|
| How you build | Assistants and Squads, set up in the dashboard or through the API. Visual Workflows retired in August 2026 | No-code visual editor with Start, Speak, Transfer and End nodes, outcomes and actions | Visual workflow builder, plus an API and MCP (Model Context Protocol) server for building in code |
| Models | Your own models for any part of the pipeline, or Vapi's providers at their own price | GPT, Claude or Gemini. Your own model on Enterprise. Thoughtly voices, Cartesia included, or ElevenLabs with your own paid key | Your own keys for any supported provider, or open-weight models on your servers |
| Telephony | A free US number for inbound, numbers imported from carriers such as Twilio and Telnyx, or a SIP (Session Initiation Protocol) trunk | Numbers bought from Thoughtly on Flex and Enterprise, or imported from Twilio or Telnyx | Your own carrier, such as Twilio, Telnyx, Plivo or Vonage |
| Pricing model | $0.05/min hosting plus each provider. Optional Core $29/mo or Pro $999/mo minimum | Flex, Scale and Enterprise quoted by sales. Unlimited minutes inside a cap of 10, 100 or more, or custom concurrent calls | Self-hosted: no per-minute platform fee. Dograh Cloud: 1¢/min platform fee. Servers, models and phone minutes still cost money |
| Compliance | SOC 2 Type II, HIPAA and GDPR. HIPAA add-on with a BAA at $2,000/mo | SOC 2 Type II report under NDA. HIPAA with a BAA on Enterprise. GDPR-aligned | Self-hosted on your own servers, with open-weight models there to keep call audio in-house |
| Channels | Voice | Voice, SMS, email, WhatsApp and iMessage | Voice |
| Lock-in | Hosted only, no on-premise option. Assistants are rebuilt to move | Hosted only. Its MCP server reads agents but never changes them | Open code under BSD-2-Clause. Agents can be built and edited through the API or MCP server |
Building agents in code or on a canvas
Vapi is built for developers, and Thoughtly for the people who run the sales funnel. In Vapi you set up Assistants, and a Squad hands a call from one assistant to another. You can work in the dashboard, but most production teams drive it through the API and their own tool servers. Vapi retired its visual Workflows on 18 August 2026, so a team that wants to draw a call flow will not find a canvas there.
Thoughtly is a canvas from the start. It has Start, Speak, Transfer and End nodes, with outcomes for branching and actions for lookups and booking. A Speak node set to Verbatim says a line exactly as written, which is handy for an opening the business must get right.
Thoughtly's own Vapi comparison frames the split as engineers against the team running the funnel. That holds for building. Some of its other cells do not match what Vapi offers today, which the compliance section below covers.
Price: a minute fee or a quoted plan
The two price in different units, so put your own volumes into both. Vapi charges $0.05 a minute for hosting, and speech-to-text, the language model, the voice and the phone line are billed on top at each provider's price. Without a package you get 4 concurrent calls, with extra lines at $10 a month each.
That fee adds up at volume. At 1,000 minutes a day, Vapi's hosting fee alone comes to about $18,250 a year, before models, phone lines or a package. Our breakdown of Vapi pricing adds up every line for three example businesses.
Thoughtly offers Flex, Scale and Enterprise through its sales team, and all three include unlimited minutes within a cap on concurrent calls. Its own comparison lists the platform fee as bundled, so the model and standard voices sit inside the quote. That makes the monthly bill easier to predict once you have it, and harder to estimate before a sales call. Our post on Thoughtly pricing works through one month for a lead team.
Compliance, hosting and lock-in
Both vendors hold the main certifications, at different prices. Vapi's FAQ says it is SOC 2 Type II-attested, HIPAA compliant and GDPR compliant, and the HIPAA add-on comes with a business associate agreement (BAA). So Thoughtly's comparison cells that tell Vapi users to build SOC 2 and HIPAA themselves do not match what Vapi offers.
Thoughtly lists a SOC 2 Type II report under NDA, HIPAA with a BAA on Enterprise and GDPR alignment, with hosting on AWS in four regions. Data residency and dedicated infrastructure also come on Enterprise. For a small team with patient calls, that means Vapi's add-on or Thoughtly's top plan.
Neither runs on your own servers. Vapi says plainly that it does not support on-premise deployments, and Thoughtly runs as a hosted service. Leaving either one means rebuilding the agent on the new platform, and Thoughtly's MCP server only reads agents, so it cannot write them elsewhere. Phone numbers are easier to keep. Numbers imported from Twilio or Telnyx stay in your own carrier account on both platforms.
Open Source Alternative to Vapi / Retell
Self-hosted voice agent platform — no per-minute fees
dograh-hq/dograh
Star on GitHub
Where an open-source platform fits
If you like one side's style but want to own the stack, an open-source platform covers both. Dograh's code is open under BSD-2-Clause, and you can run it on your own servers or on Dograh Cloud.
It offers both ways of building. A visual workflow builder suits the team that likes Thoughtly's canvas, and an API and MCP server suit the team that builds in code like Vapi's. Every model is your choice, with your own keys for any supported provider or open-weight models on your servers. Recorded clips can play your best rep's voice, with a cloned voice for anything outside them.
Phone numbers stay with your own carrier, among the telephony providers it supports such as Twilio, Telnyx, Plivo and Vonage. Each call has a trace you can read turn by turn, and QA scores each step of every call you sample.
Self-hosted Dograh has no per-minute platform fee, but servers, model keys and phone minutes still cost money. Dograh Cloud charges a 1¢-a-minute platform fee. Our guide to Vapi alternatives compares it with other platforms, and our post on how it compares with Thoughtly builds the same agent on both.
Join the Dograh Community
Dograh is an OSS alternative to Vapi. Join our Slack community for queries, releases, best practices & community interactions.
Glossary
- Hosting fee
- Vapi's $0.05-a-minute charge for running the call on its platform. Speech-to-text, the language model, the voice and the phone line are billed on top of it at each provider's price.
- Squads
- Vapi's way of joining several assistants into one call, so one assistant can hand the caller to another with a different prompt, tools or voice.
- Verbatim line
- A line in a Thoughtly Speak node that the agent says word for word, with no AI rewording. It suits openings and any wording a business must get right.
- Business associate agreement (BAA)
- The contract HIPAA requires before a vendor handles patient data for a healthcare provider. Vapi includes one with its $2,000-a-month HIPAA add-on, and Thoughtly offers one on Enterprise.

