Dograh

Bland AI vs LiveKit: A Hosted Dialer or an Open Framework

Bland AI vs LiveKit: A Hosted Dialer or an Open Framework
ComparisonNovember 17, 2026·9 min read

Bland AI vs LiveKit: A Hosted Dialer or an Open Framework

Vemu Sandeep
Vemu Sandeep·GTM Engineer, Dograh AI

GTM engineer at Dograh working across voice AI and enterprise AI. IIT (BHU) and IIM Visakhapatnam alumnus.

Bland AI suits teams that want hosted outbound calling, with its own models in $0.14 or $0.12 a minute. LiveKit suits engineers who build the dialer around open Apache-2.0 code. For teams that want a ready phone agent platform without building the dialer, Dograh is the open-source AI voice agent platform you can self-host or run on Dograh Cloud.

Key Takeaways

  • Bland's Start plan stops at 100 calls a day, so real campaigns need Build.
  • LiveKit's own numbers take inbound calls only, so outbound needs your SIP trunk.
  • At 15,000 minutes, Bland Build costs $2,099 with models, LiveKit Ship $190 before them.

Bland AI and LiveKit can both run an outbound calling operation, but one sells it as a hosted product and the other hands your engineers the code. We compare caps, self-hosting, models and a priced month, then show where open-source Dograh fits when you want both.

Bland AI and LiveKit in one table

The two sit at different layers, so some rows set a finished feature against the code you would write to get it.

Bland AILiveKit
What it isHosted voice agent platformOpen-source agent framework and media server, plus LiveKit Cloud
CodeClosedApache-2.0, turn detector under the LiveKit Model License
Who builds the agentYour team, or Bland's forward deployed engineers with your ops teamYour engineers, in Python or Node
ModelsBland's own, included in the per-minute rateAny provider on your keys, or LiveKit Inference
Price shapeStart $0.14 a minute. Build $299 a month plus $0.12 a minuteBuild $0, Ship from $50, Scale from $500, then $0.01 per agent minute. Models extra
Calls at once10 on Start, 50 on Build5 on Build, 20 on Ship, up to 600 on Scale
Call caps100 a day and 100 an hour on Start. 2,000 a day and 1,000 an hour on BuildSet by sessions at once on each plan
Outbound callingBland's carrier at pass-through, or your ownYour own SIP trunk, since LiveKit's numbers are US and inbound only
Self-hostingBland's closed product in your cloud or on your premises, on EnterpriseThe open code on your servers, no contract needed
HIPAABAA included on EnterpriseSigned BAA on Scale and Enterprise

Running an outbound campaign on each

Outbound is where Bland is at home, and its plan limits shape the campaign before the per-minute rate does. Bland's Start plan has a $0 platform fee, 10 calls at once, 100 calls a day and 100 an hour. Build lifts that to 50 at once, 2,000 a day and 1,000 an hour for $299 a month. Each call starts with one API request, and a transfer to a person costs $0.05 a minute on Start and $0.04 on Build, or nothing extra on your own telephony.

Phone minutes sit outside Bland's rate. They are billed on Bland's carrier at pass-through cost or on your own carrier, so the dialing side of a campaign stays simple to set up.

LiveKit starts the other way round. LiveKit Phone Numbers are available in the US only and take inbound calls, with outbound support listed for a future release. An outbound agent on LiveKit therefore dials through a SIP trunk from a carrier you sign with. LiveKit Cloud's Ship plan allows 20 agent sessions at once, which a burst campaign can reach quickly, and Scale starts at $500 a month.

The bigger gap is the product around the agent. Contact lists, calling hours, retry rules, a screen where someone reads transcripts, and QA on live calls are code your engineers write against LiveKit's APIs. LiveKit's Agent Builder suits browser prototypes, and anything with handoffs or tests moves to code. Our comparison of Retell AI vs LiveKit walks through that build list in more detail.

What self-hosting means for each

Both vendors now use the word self-host, and they mean different things by it. Bland says you can run it in your own cloud, on your premises, or in Bland's cloud, and its Enterprise plan lists on-prem or VPC deployment. The code stays closed. What you get is Bland's product running inside your environment under an Enterprise contract, with Bland's own models and infrastructure behind every call.

LiveKit's version is open code. LiveKit Agents on GitHub carries the Apache-2.0 license, so a team can run the framework and media server on its own servers without asking anyone. The trade is what stays behind in LiveKit Cloud. Phone numbers, LiveKit Inference and the Agent Builder do not come with the open code, so a self-hosted install also needs its own SIP service and every model key. LiveKit's turn detection models sit under a separate LiveKit Model License, which teams with a strict open-source policy should read first.

Models in the rate or models you choose

Bland keeps the whole model stack in house. Bland says every call runs on its own models and infrastructure with no third-party vendors. There are no token charges, and transcription and premium voices sit inside the rate. That bundled rate makes the bill easy to forecast. It also means the plan rate is the only lever on that part of the cost, because there is no other language model or voice provider to switch to.

LiveKit leaves every model choice to you. Agents connect to speech-to-text, language model and voice providers through plugins on your own keys, or through LiveKit Inference on one bill. On Build and Ship, LiveKit Inference charges $0.0030 a minute for GPT-5.4 mini, $0.0048 for Deepgram Nova-3 and $0.0300 for Cartesia Sonic 3. You can tune each part, but someone has to pick it, test it and watch its limits.

Dograh

Open Source Alternative to Vapi / Retell

Self-hosted voice agent platform — no per-minute fees

dograh-hq/dograh

Star on GitHub

What 15,000 outbound minutes cost

One priced month shows how differently the two bill. Take an example of 15,000 outbound minutes, about 5,000 calls of 3 minutes, or roughly 165 calls a day. Bland's Start plan stops at 100 calls a day, so this volume needs Build.

LineBland AI BuildLiveKit Cloud Ship
Plan$299 a month$50, with 5,000 agent minutes
Platform minutes15,000 x $0.12 = $1,80010,000 extra agent minutes x $0.01 = $100
SIP minutesCarrier minutes billed separately10,000 past the 5,000 included x $0.004 = $40
ModelsIncludedOne example stack on LiveKit Inference, 15,000 x $0.0378 = $567
Total$2,099 plus carrier minutes$190 before models, about $757 with that stack, plus carrier minutes

The rows are not like for like. Bland's $2,099 already pays for transcription, the language model and the voice. LiveKit's $190 covers agent and SIP minutes only, and its $757 still leaves out the dialer logic, the review screens and the engineers who build and run them. Our breakdowns of Bland AI pricing and LiveKit pricing go through each plan line by line.

Join the Dograh Community

Dograh is an OSS alternative to Vapi. Join our Slack community for queries, releases, best practices & community interactions.

Which one fits your team

Bland fits an operations team that wants campaigns running in weeks and has no engineers to spare for a dialer. The caps are clear, the models come included and Bland's engineers can build with you. Plenty of teams never need to leave a platform like that. The cost is closed code, a model stack you cannot change and plan limits that set your daily volume.

LiveKit fits engineers who want the agent as code and expect to go past any platform's presets. You choose every model, run the open code where you like and pay little for the platform itself. The cost is the campaign layer, from the SIP trunk to retries and call review, which your team builds and maintains.

A team with an outbound campaign has two options. Bland AI stands as a finished hosted product with the models included and Bland's engineers beside the agent. On the LiveKit lot, your engineers stack the campaign layer by hand beside the agent, from the SIP trunk up through contact lists, calling hours, retry rules, call review and QA on live calls.
Bland fits an operations team that wants campaigns running in weeks and has no engineers to spare for a dialer. LiveKit fits engineers who want the agent as code and expect to go past any platform's presets.

Whichever way you lean, test with real outbound calls before you commit. Most phone calls still reach the agent as 8 kHz audio while a browser demo runs at 16 kHz, so an agent that sounds right in the dashboard can mishear names on a real line.

Where Dograh fits for open-source and on-prem teams

Some teams want LiveKit's open code with the campaign layer Bland ships. Dograh, the open-source AI voice agent platform, is built on a fork of Pipecat under the BSD 2-Clause license, and new agents start in the Agent Builder, where you describe the agent and "Approve & build" builds it for review. A self-hosted install needs a Dograh Service Key for the builder, or you build on the canvas. For outbound lists, Campaigns dial a contact file with calling hours by weekday in one time zone, and retry busy or unanswered numbers up to 10 times, 2 by default.

You pick the speech-to-text, language model and voice provider for each agent, on your own keys or open models you host, and bringing your own key moves the contract while the audio still reaches that provider. When a contact asks for a person, Transfer Call works on Twilio, Telnyx, Plivo, Cloudonix and Asterisk ARI. Dograh calls the destination first and connects the contact only when someone answers, and a busy, failed or unanswered transfer comes back to the agent so the call carries on.

Where it runs is your call, and so is the cost that comes with it. You can self-host with Docker on at least 8 GB of RAM and 4 vCPUs, use the managed cloud, or take on-prem deployment on Enterprise. A self-hosted install runs 10 calls at once by default, a setting you can raise, and Cloud's pay as you go plan includes 10, with more on request. 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, $150 for the 15,000-minute month above. Running it yourself means the server, the database and the upgrades are your team's job, with community support on the open-source edition and custom SLAs on Enterprise.

After each call, the QA node scores every step of test and live calls on a sample rate you set, and View Trace shows the speech-to-text, AI and voice time for each turn. If LiveKit is your starting point, our post on the open-source alternative to LiveKit compares the two self-hosted installs part by part.

Glossary

Call cap
A plan limit on how many calls can start in one day, separate from how many run at once. Bland's Start plan allows 100 a day and Build 2,000.
SIP trunk
A SIP connection from a phone carrier that lets an agent place calls. LiveKit's own numbers take inbound calls only, so outbound on LiveKit runs through one.
Agent minute
LiveKit Cloud's billing unit for time an agent spends in a session. Ship includes 5,000 a month and charges $0.01 a minute after that.
Bundled rate
A per-minute price that already includes speech-to-text, the language model and the voice. Bland's $0.14 and $0.12 rates work this way, with phone minutes billed apart.

Frequently Asked Questions

Get started with Dograh

Build, deploy, and scale AI agents with Dograh. Join the community of developers building the future.