Gradium says its text to speech starts in under 50ms, against the 199ms Bland publishes, and a team building a voice agent on latency alone will take that number seriously. Gradium sells it with five languages, monthly plans whose credits don't roll over, and a price per million characters that runs from $36 to $58 depending on the plan.
Bland Speech v3 scored 1365 in Design Arena's blind listening test of 4 August 2026, and Gradium wasn't in those results. So the comparison turns on speed, price, languages, cloning, and where each model can run.
Which is better, Bland Speech or Gradium?
For a voice agent on a phone line where callers should hear a person, Bland Speech is the better choice as of November 2026. Bland Speech v3 scored 1365 in Design Arena's Audio Realism Bench results of 4 August 2026, and Bland bills pay as you go with credits that do not expire, where Gradium's plan credits end each month.
Gradium makes more sense for a team that puts first-audio latency first, speaks English, French, German, Spanish, or Portuguese, or wants dedicated instances, self-hosting, or on-premise deployment from Gradium.
Use Bland Speech when
- Callers on a phone line should hear a voice that sounds like a person
- A team wants pay as you go billing with credits that do not expire, and no monthly allowance to use up
- An existing OpenAI audio speech integration should switch providers by changing a base URL, a key, and a model name
- A script needs performed delivery, such as a laugh or a warm read, written inline as tags
- A voice agent runs on Pipecat or LiveKit Agents and needs a text to speech service built for those frameworks
Use Gradium when
- A voice agent puts first-audio latency first and will test Gradium's under 50ms claim on its own traffic
- A team speaks English, French, German, Spanish, or Portuguese and wants mid-sentence switching between them
- A team spends enough on speech that a $340 or $1,615 monthly plan lowers the price per character
- A team wants guides for Vapi, Pipecat, and LiveKit Agents, or migration guides from ElevenLabs, Cartesia, Deepgram, or Fish Audio
- A security team wants dedicated instances or an on-premise deployment from Gradium
What these rows mean
- Audio Realism Bench, 4 August 2026: Audio Realism Bench is Design Arena's blind listening test, where native speakers hear two clips of the same line from two models and pick the one that sounds more like a real person. A higher score means listeners picked that voice more often.
- Price per million characters: Bland Speech costs $0.04 per 1,000 characters, which is $40 per million. Gradium bills one credit per character, and its price per million credits is $58 on XS, $48 on S, $38 on M, and $36 on L, worked out on monthly billing.
- Monthly plan: Gradium shows monthly-billing prices, and shows lower prices with one month free when billed yearly. Plan credits must be used within the month and cannot roll over.
- Free usage: New Bland accounts get 50,000 characters free, nearly an hour of speech. Gradium's Free plan includes 45,000 credits a month, and commercial use starts at XS.
- First audio: Each is the company's own published claim. Bland publishes 199ms to first audio on a streamed request. Gradium's product page says "under 50ms" and cites 49ms at the median (P50) on a Speko benchmark.
How do Bland Speech and Gradium compare?
Both companies build for a team running a voice agent. Gradium sells speed first, through monthly credit plans, and Bland built Bland Speech for phone calls, where in Bland's words polish reads as fake, and bills the characters a team uses.
Bland Speech overview
Bland Speech is a closed-source model Bland built itself, the text to speech model inside Bland, a voice AI provider that runs its own model and speech for enterprise phone calls. Bland Speech is sold on its own as an API and a studio on the Bland Speech page.
A team sends text over HTTP to POST /v2/tts and gets raw audio or a WAV file back, or streams language model tokens into a WebSocket at WSS /v2/tts/ws, where a new context interrupts the turn that is playing. POST /v2/audio/speech follows the OpenAI audio speech format, and a first-party Pipecat service and a LiveKit Agents plugin for Python are available. The text to speech API page has the code.
Pros
- Bland Speech v3 ranked 1st of 13 AI voices on Audio Realism Bench on 4 August 2026, at 1365
- Pay as you go at $0.04 per 1,000 characters, with no monthly plan and credits that do not expire
- An OpenAI-compatible endpoint, so an existing integration moves over by changing a base URL
- Performance tags, and a director in the studio for how a line is read
- Mu-law at 8 kHz and PCM from 8 to 48 kHz, ready for phone calls
- Instant clones from about 10 seconds of audio, and professional clones from 30 minutes or more
Cons
- More than a quick prototype needs, because Bland Speech is built for production calls
Gradium overview
Gradium sells real-time voice AI for voice agents: text to speech, speech to text, live translation, voice cloning, and voice design. A WebSocket takes text token by token, as a language model writes it, and a REST call handles complete text.
Gradium's API calls its text to speech model default, and Gradium updates that model in place, most recently on 9 November 2026 to handle break durations more accurately. Gradium supports five languages, English, French, German, Spanish, and Portuguese, each with regional variants on every plan, and says its models switch languages mid-sentence. Gradium also publishes Phonon, small open-weight models for phones, laptops, and browsers.
Pros
- A first-audio claim of under 50ms, with 49ms at the median on a Speko benchmark
- Price per million credits falls to $38 on M and $36 on L, below Bland Speech's $40
- Mu-law and A-law output at 8 kHz for phone lines
- Instant clones from 10 seconds of audio, and professional clones from 30 minutes to a few hours
- ISO 27001:2022, SOC 2 Type 2, and HDS v2 on Gradium's Trust Center, and a GDPR data processing agreement
- Guides for LiveKit Agents, Pipecat, and Vapi, and migration guides from four rivals
Cons
- Plan credits must be used within the month and cannot roll over
- $58 per million credits on XS and $48 on S, above Bland Speech's $40
- Commercial use starts on the XS plan and not on Free
- Professional clones carry a one-time training fee of 1M credits each, then 1.2 credits per character
How do Bland Speech and Gradium handle phone audio and latency?
Gradium publishes the lower first-audio number, and Gradium's own two numbers differ. Gradium says its text to speech starts in under 50ms and cites 49ms at the median on a Speko benchmark and 50ms on Coval, while Gradium also says its expected time to first token is below 300ms when streaming from servers in Europe and the US. Bland publishes 199ms to first audio on a streamed request.
Bland Speech outputs mu-law at 8 kHz and PCM at 8, 16, 24, 44.1, or 48 kHz. Gradium outputs mu-law and A-law at 8 kHz, PCM at 8, 16, 22.05, 24, 44.1, and 48 kHz, WAV at 48 kHz, and Opus, and publishes a telephony recipe that uses mu-law or A-law for audio going out to the phone network.
How do Bland Speech and Gradium handle security?
Gradium holds ISO 27001:2022, SOC 2 Type 2, and HDS v2, a French health data hosting certification, and names Scaleway, Amazon Web Services, and, from 9 December 2026, Google Cloud Platform as cloud providers. Gradium says teams under HIPAA or a strict internal policy "usually take the VPC route", and sells dedicated instances, self-hosting, and on-premise deployments.
Bland is SOC 2 Type II audited, HIPAA compliant with a signed BAA, and FedRAMP 20x Class C certified at Moderate, granted on 17 November 2026. Speech synthesis runs inside that FedRAMP boundary, and Bland runs its own model and speech with zero AI sub-processors in the call path, as Bland's trust and security record sets out. Enterprise customers get dedicated orchestration and dedicated GPUs run by Bland.
How much do Bland Speech and Gradium cost?
Gradium costs more than Bland Speech per character on its two smallest plans and less on its two largest. Bland Speech costs $40 per million characters, and Gradium's price per million credits is $58 on XS, $48 on S, $38 on M, and $36 on L.
Bland Speech pricing
Bland Speech costs $0.04 per 1,000 characters, directed or not, in the studio or the API. New accounts get 50,000 characters free, and credits do not expire. A 60-second note costs about $0.04, a 10-minute narration about $0.36, and a million characters $40.
Enterprise pricing is custom, set with Bland's sales team.
Gradium pricing
Gradium bills one credit per text to speech character, about 45,000 characters to an hour of speech. The monthly-billing prices are Free at $0 with 45,000 credits, XS at $13 with 225,000, S at $43 with 900,000, M at $340 with 9 million, and L at $1,615 with 45 million, and billing yearly gives one month free. Credits from a plan can't roll over to the next month, and pay as you go credits do.
What does a month of speech cost on each?
At 100,000 characters a month, about two hours of speech, Bland Speech costs $4 and Gradium's XS plan costs $13, with 225,000 credits. At 900,000 characters, about 20 hours, the figures are $36 and $43 on S. At 9 million characters, about 200 hours, Bland Speech costs $360 and Gradium's M plan costs $340. At 45 million characters, about 1,000 hours, Bland Speech costs $1,800 and Gradium's L plan costs $1,615.
Those Gradium totals assume a team uses a plan's full allowance every month, because unused plan credits expire.
Which sounds more human, Bland Speech or Gradium?
Design Arena's blind tests of 4 August 2026 scored Bland Speech v3 at 1365, second only to a recording of a real person, and Gradium wasn't among the 13 AI voices scored.
- #1BlandBland Speech v31365
- 2ndMicrosoftMAI-Voice-21224
- 3rdxAIGrok TTS1156
- 4thOpenAIGPT Realtime 21119
- 5thGoogleGemini 2.5 Pro TTS1109
- Not rankedGradium's own voicesNot ranked
Ranks are among 13 AI voices, and Gradium's own voices weren't ranked. Real human recordings scored 1500, at the right edge of this scale. Bars start at 600, not zero. Source: Design Arena's published results.
Design Arena published how listeners heard two clips of the same line, without knowing which model made which, in an article on Audio Realism Bench. Gradium publishes a test of its own, an 81.0% pass rate on 500 hard sentences in five languages, which it sets against 75.1% for Cartesia Sonic 3.6 and 65.4% for ElevenLabs v3. That is a test of hard text such as numbers, emails, and customer IDs, run by Gradium, and a different question from which voice sounds more like a person.
Why is Bland Speech the best Gradium alternative?
A team that chose Gradium for speed and now wants a voice judged on realism can compare it with Bland Speech v3, which scored 1365 in the blind tests of 4 August 2026 and returns mu-law at 8 kHz without a monthly credit plan to manage.
Bland Speech credits do not expire, so a quiet month costs nothing extra, where Gradium plan credits expire at the end of each month. A team on OpenAI's audio speech API can switch by changing a base URL, and both companies clone a voice from about 10 seconds of audio, with Bland's professional clones going through a consent and rights review and staying private to the organization that made them.
Bland's comparison of text to speech APIs covers more vendors, and the head-to-head comparisons collect the pairs. A team that wants Gradium on its own can read the Gradium review or the Gradium alternatives, and one comparing speed claims can start with time to first audio.
Frequently asked questions
Is Bland Speech better than Gradium?
For phone calls and voice agents where the voice should sound like a person, Bland Speech is better than Gradium. Bland Speech v3 scored 1365 on Audio Realism Bench on 4 August 2026, and Gradium wasn't in those results. Gradium suits a team that wants the lower published first-audio figure of the two.
Is Gradium cheaper than Bland Speech?
That depends on the plan. Gradium's price per million credits is $58 on XS and $48 on S, above Bland Speech's $40, and $38 on M and $36 on L, below it. Plan credits expire each month, so those prices hold when a team uses its whole allowance.
Which is faster, Bland Speech or Gradium?
On published claims, Gradium. Gradium says text to speech starts in under 50ms, and Bland publishes 199ms to first audio on a streamed request. Gradium also says its expected time to first token is below 300ms when streaming, and each figure comes from its own company's method.
Which languages does Gradium support?
Gradium supports five languages: English, French, German, Spanish, and Portuguese, each with regional variants on every plan. Gradium's Phonon 40m multilingual open-weight model covers the same five.
Does Gradium clone voices?
Yes. Gradium clones a voice from 10 seconds of clear audio, and builds a professional clone from 30 minutes to a few hours of audio on the M plan or higher. A professional clone costs a one-time training fee of 1M credits, then 1.2 credits per character.
Does Gradium work with LiveKit, Pipecat, and Vapi?
Yes. Gradium has guides for LiveKit Agents, Pipecat, and Vapi, and migration guides from ElevenLabs, Cartesia, Deepgram, and Fish Audio. Bland Speech has a first-party Pipecat service and a LiveKit Agents plugin for Python.
Does Bland Speech work on phone calls?
Yes. Bland built Bland Speech for phone calls, and it returns mu-law at 8 kHz as well as PCM from 8 to 48 kHz, so audio goes onto a phone line without conversion.
