ElevenLabs and Cartesia both sell speech by the plan, with a monthly allowance of characters or credits to manage, while Bland Speech charges $0.04 for every 1,000 characters and has no plan to pick. That difference shapes most of what a team pays and keeps an eye on each month, so it comes before the question of how each voice sounds.
A phone agent also lives or dies on whether the caller hears a person, and both rivals have shipped newer models since the independent listening test that included them. The text to speech comparison hub lists every other provider.
Which is best for a phone call: Bland Speech, ElevenLabs, or Cartesia?
For a phone line, Bland Speech is the pick of the three as of November 2026. Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, it was built for phone calls, and it bills one standard per-character rate with no plan. ElevenLabs is the pick for a team that needs the widest language range, with 90+ languages across its Eleven v4 family, and Cartesia is the pick for the smallest first-audio figure of the three, under 90ms.
What these rows mean
- Audio Realism Bench: Design Arena's blind listening test. A listener hears two clips of the same line from two models and picks the one that sounds more like a real person, and the votes become an Elo rating, a score that rises each time a voice is picked.
- Price per million characters: Cartesia's plans bill 1 credit per character, which works out to $37.38 on Scale ($299 for 8 million credits) up to $50 on Pro ($5 for 100,000 credits). ElevenLabs' $0.04 and $0.08 per 1,000 characters depend on the model.
- First audio, as published: Each number is the company's own claim, and none is a race Bland ran.
- Phone audio: Mu-law is 8-bit compressed audio at 8 kHz, used on North American and Japanese phone lines and by Twilio. A-law is the other common phone encoding.
How fast does each one start talking?
Each company publishes its own first-audio number, and the three count different things. Time to first audio is the gap between sending text and receiving the first sound, and a voice agent's reply can't begin before it. Bland publishes 199ms on a streamed request, counted to the first audio chunk rather than the finished file. Cartesia says Sonic delivers first audio in under 90ms, which it describes as model latency. ElevenLabs puts the median for Eleven v4 Turbo at about 100ms excluding application and network latency, and also states a median time to first speech of about 150ms.
On those numbers Bland Speech starts slowest of the three, by more than 100 milliseconds against Cartesia's figure. A voice agent also waits on speech recognition and a language model before any voice is generated, so text to speech is one part of the delay a caller hears, and a team that needs the smallest figure should count Cartesia's as a point in its favor.
What does each send down a phone line?
All three return audio a US phone line can carry without conversion. Bland Speech returns mu-law at 8 kHz, or PCM at 8, 16, 24, 44.1, or 48 kHz, and v3 renders natively at 48 kHz. Cartesia returns mu-law, A-law, and PCM, and ElevenLabs returns ulaw_8000, alaw_8000, and pcm_8000 among its output formats.
Where do ElevenLabs and Cartesia fall short next to Bland Speech?
ElevenLabs sends its newest models down a different route than its standard streaming connection, and Cartesia meters speech in credits that block new requests when they run out and overages are off.
Where ElevenLabs falls short
ElevenLabs sells Eleven v4 and Eleven v4 Turbo, launched on 28 September 2026, and calls v4 Turbo its state of the art model for real-time speech. The catch is the route in. A WebSocket is a connection that stays open so text can arrive in pieces as a language model writes it, and ElevenLabs' standard text to speech WebSocket doesn't support Eleven v3 or v4. Eleven v4 runs through ElevenLabs' Text to Dialogue API and v4 Turbo through its Text to Dialogue WebSocket, so a team moving an existing integration to the new models changes endpoints as well as model names.
Price comes in two steps, $0.08 per 1,000 characters on v4 and $0.04 on v4 Turbo. Free accounts may use ElevenLabs for non-commercial purposes only, and the commercial license starts on the $6 Starter plan.
Pros
- 90+ languages across the Eleven v4 family
- Audio tags such as [whispering] and [laughing] on Eleven v4 and v3
- A voice captcha check before a professional clone is built
- Mu-law and A-law output at 8 kHz for phone lines
- Text to speech can run inside a customer's own cloud under a private deployment
Cons
- Eleven v3 ranked 12th of 13 AI voices on Audio Realism Bench on 4 August 2026, and Eleven v4 was released after those results
- Two prices for the newest models, $0.08 and $0.04 per 1,000 characters
- Professional clones need the Creator plan or above
Use ElevenLabs when:
- A team ships agents in dozens of languages and will work from one vendor's 90+ language family
- A team scripts delivery with audio tags and will route Eleven v4 through Text to Dialogue to use them
Where Cartesia falls short
Cartesia sells Sonic 3.6, released on 27 August 2026, and bills 1 credit per character from monthly plans: Free at $0 with 20,000 credits, Pro at $5 with 100,000, Startup at $49 with 1.25 million, and Scale at $299 with 8 million. The credit meter has edges a phone team hits. Each break tag in a transcript costs 1 credit, adding an accent to a voice costs 225 credits, and with overages switched off, new requests are blocked once credits run out.
Cartesia's emotion controls are in beta and work for English only. Speed and volume controls don't apply to a professional voice clone, which learns pacing and loudness from its training audio.
Pros
- 44 languages from one model, Sonic 3.6
- First audio in under 90ms, as Cartesia publishes it
- Word-level timestamps on the WebSocket and SSE endpoints, and pronunciation dictionaries
- Mu-law and A-law output at 8 kHz for phone lines
- On-premise, air-gapped, and own-VPC deployment under enterprise contracts
Cons
- Sonic 3.5 ranked 6th of 13 AI voices on Audio Realism Bench on 4 August 2026, and Sonic 3.6 was released after those results
- New requests are blocked at zero credits when overages are off
- The Free plan is for evaluation and personal use, and the commercial license starts on Pro
Use Cartesia when:
- A team builds on 44 languages from one model and will manage a credit allowance each month
- An engineering team wants word-level timing from a WebSocket to line captions or visuals up with speech
What do Bland Speech, ElevenLabs, and Cartesia cost?
One million characters of speech costs $40 on Bland Speech, $40 or $80 at ElevenLabs' API rates depending on the model, and between $37.38 and $50 inside a Cartesia plan, with Cartesia's overage rates running from $38 to $65 per million credits.
Bland Speech pricing
Bland Speech costs $0.04 per 1,000 characters, which is $40 per million, pay as you go and the same in the studio and the API, with no monthly plan. A new account gets 50,000 characters free once, and credits don't expire. Some public-library voices add a per-character creator fee on top. Enterprise pricing is custom, set with Bland's sales team.
ElevenLabs pricing
ElevenLabs bills its API in US dollars: $0.08 per 1,000 characters on Eleven v4, Eleven v3, and Multilingual v2, and $0.04 on Eleven v4 Turbo, v3 Conversational, and Flash. Plans run Free, Starter at $6, Creator at $22, Pro at $99, Scale at $299, and Business at $990 a month, with Enterprise custom, and each plan's included characters double at the Turbo rate.
Cartesia pricing
Cartesia's plans are monthly, at 1 credit per character. Overage is $65 per million credits on Pro, $45 on Startup, and $38 on Scale, billed at renewal and only when switched on. Enterprise is custom.
What does a month of speech cost on each?
Worked from the published prices, with no tax or phone charges counted:
- 100,000 characters: Bland Speech $4, Cartesia's Pro plan $5, and ElevenLabs' Starter plan $6, which covers 150,000 characters at the Turbo rate.
- 1 million characters: Bland Speech $40. ElevenLabs $40 on v4 Turbo or $80 on v4 at the API rates. Cartesia's Startup plan covers it at $49, and 1 million characters on Pro costs $63.50, which is the $5 plan plus 900,000 characters of overage at $65 per million.
- 10 million characters: Bland Speech $400. ElevenLabs $400 on v4 Turbo or $800 on v4. Cartesia's Scale plan comes to $375, which is $299 for 8 million credits plus 2 million at $38 per million.
At 10 million characters Cartesia's Scale plan beats Bland Speech's per-character rate, and a team with steady volume can price a plan against that rate with these numbers. The case for Bland Speech rests on how it sounds and on having no allowance to run out.
Which sounds more human, Bland Speech, ElevenLabs, or Cartesia?
Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, at 1365, where the human recording scored 1500. Cartesia's Sonic 3.5 ranked 6th at 1096 and ElevenLabs' Eleven v3 ranked 12th at 938. Design Arena describes the result in its own words: "Second only to human speech is Bland Speech v3 by Bland AI, with an Elo of 1365."
- #1BlandBland Speech v31365
- 2ndMicrosoftMAI-Voice-21224
- 3rdxAIGrok TTS1156
- 4thOpenAIGPT Realtime 21119
- 5thGoogleGemini 2.5 Pro TTS1109
- 12thElevenLabsEleven v3938
Ranks are among 13 AI voices, and ElevenLabs's own voice is on the list. Real human recordings scored 1500, at the right edge of this scale. Bars start at 600, not zero. Source: Design Arena's published results.
Two limits apply. The test uses one female and one male voice per model, chosen as its most naturally conversational American-English voice, so a team working in other languages or long narration should listen for itself. And both rivals released newer models after those results, Sonic 3.6 on 27 August 2026 and Eleven v4 on 28 September 2026, so the 4 August placings say nothing about how those compare.
Why is Bland Speech the pick over ElevenLabs and Cartesia for phone calls?
Bland Speech is the pick over ElevenLabs and Cartesia when the product is a phone call, because Bland Speech v3 was built for calls and the bill has no allowance, credit meter, or plan tier to manage. Bland runs its own model and speech instead of wrapping a third party's.
A team sends text over HTTP to /v2/tts and gets raw audio or a WAV file back, or streams a language model's output into a WebSocket at /v2/tts/ws, where a new context interrupts the turn that is playing, so barge-in is part of the protocol. Performance tags in square brackets, such as [laughs], [sighs], or [say warmly], shape how the words after them are delivered. The endpoints are covered under the text to speech API, and Bland Speech has the studio.
Bland, the company behind the model, is SOC 2 Type II audited, HIPAA compliant with a signed BAA, and FedRAMP 20x Class C certified at Moderate, granted 17 November 2026, and its trust page says speech synthesis runs inside the FedRAMP boundary.
Pros
- Ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026
- One standard rate of $0.04 per 1,000 characters, with no plan and no credit meter
- Mu-law and PCM output, from 8 to 48 kHz
- Performance tags, plus a WebSocket where a new context interrupts a turn
- A first-party Pipecat service and a LiveKit plugin
- An instant clone from about 10 seconds of audio
Cons
- More than a hobby project needs, because Bland Speech is built for production calls
Bland Speech vs ElevenLabs and ElevenLabs vs Cartesia cover the two-way versions, and the Cartesia alternatives and ElevenLabs migration guide serve a team already on a rival. Every provider in the series is on the rival comparisons page.
Frequently asked questions
Is Bland Speech better than ElevenLabs or Cartesia?
For phone calls, Bland Speech is the stronger pick: Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, ahead of Cartesia's Sonic 3.5 and ElevenLabs' Eleven v3. ElevenLabs covers more languages and Cartesia publishes a smaller first-audio figure, so the better pick depends on which a team needs most.
Which is cheapest for 1 million characters a month?
Bland Speech and ElevenLabs' v4 Turbo both cost $40 per million characters at published rates, and Cartesia's Startup plan covers 1 million characters for $49. ElevenLabs' Eleven v4 costs $80 per million. At 10 million characters Cartesia's Scale plan comes to $375 against $400 for Bland Speech.
Can each one clone a voice?
Yes, each with different requirements. Bland Speech makes an instant clone from one clean sample of about 10 seconds and a professional clone from 30 minutes or more of audio after Bland's consent and rights review, and every clone is private to the customer's organization, with the customer confirming the voice is theirs to use. Cartesia's instant clone starts on Pro and its professional clone on Startup, while ElevenLabs' start on Starter and Creator, with a voice captcha check for professional clones.
Do they work with Pipecat and LiveKit?
Yes. Pipecat's list of text to speech services includes Bland, Cartesia, and ElevenLabs, each as a service the Pipecat team maintains. LiveKit Agents has plugins for all three, with Bland's in Python only and ElevenLabs' and Cartesia's in Python and Node.js.
How do the three compare on compliance for regulated calls?
ElevenLabs claims SOC 2 Type 2, HIPAA, GDPR, and PCI DSS 4.0.1 Level 1, with BAAs on Enterprise, and Cartesia says it is SOC 2 Type II certified and claims HIPAA, GDPR, and PCI DSS 4.0.1, also with BAAs on Enterprise. Bland adds FedRAMP 20x Class C certification at Moderate, granted 17 November 2026, with speech synthesis inside the FedRAMP boundary, alongside SOC 2 Type II and HIPAA compliance with a signed BAA.
