Model Index & per-minute prices
The Callable Model Index at models.callableio.com shows what one minute of a phone conversation costs on Callable, for every language model, transcriber, voice and phone line an agent can run on. It has a calculator: pick the four parts of your agent and it adds up your price per minute.
Every price on it, and on this page, is what Callable charges you, with the platform fee already included. It comes from the same price tables your credit balance is billed from, so the index, these docs and your bill always agree. Prices were last checked against the providers on 10 November 2026.
What a call is made of
Every minute of a phone call uses four things at once, and each is priced separately:
| Part | What it does | How it's priced |
|---|---|---|
| Language model (LLM) | Decides what the agent says | Per token read and written |
| Transcriber | Turns the caller's speech into text | Per minute the call is connected |
| Voice | Speaks the agent's reply | Per character spoken |
| Phone line | Carries the call to the country you dial | Per minute, by destination |
Your price per minute is the sum of the four. A web chat uses only the language model.
How a minute is priced
Language models and voices aren't priced by the minute, so the index converts them using what a typical minute of conversation on Callable uses:
- About 5 agent replies a minute. On every reply the language model re-reads the agent's instructions and the conversation so far, so it reads about 10,000 tokens and writes about 100 a minute. Reading is most of the cost.
- About 500 characters spoken by the agent a minute.
- The transcriber and the phone line run for every minute the call is connected.
Your own calls can cost more or less than the index:
- Longer instructions or a big Knowledge Base cost more, because the model re-reads them on every reply.
- A chatty agent costs more on the voice, a quiet one less.
- Short calls: the line is billed by the minute.
The Usage page always shows what your calls actually cost.
An example
A new agent starts on this stack:
| Part | Choice | Per minute |
|---|---|---|
| LLM | GPT-4o mini (OpenAI) | $0.0031 |
| Transcriber | Nova-3 (Deepgram) | $0.0086 |
| Voice | Turbo v2.5 (ElevenLabs) | $0.1500 |
| Phone line | United States | $0.0140 |
| Total | $0.1757 (18 credits) |
That's $175 for 1,000 minutes of calls. The voice is most of it. The cheapest option in each part brings the same call to $0.0361 a minute: Gemini 2.5 Flash-Lite, Universal-Streaming, S2 Pro, United States.
To try your own combination, use the calculator at the top of models.callableio.com.
Language models
Each price is for one minute of conversation; the unit price is per million tokens read and written.
| Model | Provider | Per minute | Credits / min | Unit price |
|---|---|---|---|---|
| Gemini 2.5 Flash-Lite | $0.0021 | 0.21 | $0.20 in · $0.80 out per 1M tokens | |
| GPT-6 Luna | OpenAI | $0.0021 | 0.21 | $0.20 in · $1.00 out per 1M tokens |
| GLM-5.3 Flash | Z.ai | $0.0031 | 0.31 | $0.30 in · $1.00 out per 1M tokens |
| GPT-4o mini | OpenAI | $0.0031 | 0.31 | $0.30 in · $1.20 out per 1M tokens |
| Anthropic Haiku 5.5 | Anthropic | $0.0032 | 0.32 | $0.30 in · $1.50 out per 1M tokens |
| GLM-4.5 Air | Z.ai | $0.0042 | 0.42 | $0.40 in · $2.20 out per 1M tokens |
| GPT-5.4 Nano | OpenAI | $0.0043 | 0.43 | $0.40 in · $2.50 out per 1M tokens |
| Gemini 3.1 Flash-Lite | $0.0053 | 0.53 | $0.50 in · $3.00 out per 1M tokens | |
| Gemini 3.5 Flash-Lite | $0.0065 | 0.65 | $0.60 in · $5.00 out per 1M tokens | |
| GPT-5.4 Mini | OpenAI | $0.0159 | 1.6 | $1.50 in · $9.00 out per 1M tokens |
| Anthropic Sonnet 5.5 | Anthropic | $0.0630 | 6.3 | $6.00 in · $30.00 out per 1M tokens |
| Anthropic Sonnet 5 | Anthropic | $0.0630 | 6.3 | $6.00 in · $30.00 out per 1M tokens |
Anthropic models on Callable's key include a premium over the other providers, already in these prices.
Transcribers
Transcribers listen for the whole call, so the price is per minute connected.
| Transcriber | Provider | Per minute | Credits / min | Unit price |
|---|---|---|---|---|
| Universal-Streaming | AssemblyAI | $0.0050 | 0.50 | $0.30 per hour of audio |
| Nova-3 | Deepgram | $0.0086 | 0.86 | $0.52 per hour of audio |
| Nova-2 Phone | Deepgram | $0.0116 | 1.2 | $0.70 per hour of audio |
| Nova-2 | Deepgram | $0.0116 | 1.2 | $0.70 per hour of audio |
| Scribe v2 Realtime | ElevenLabs | $0.0130 | 1.3 | $0.78 per hour of audio |
| Enhanced | Deepgram | $0.0290 | 2.9 | $1.74 per hour of audio |
| GPT-Live-Transcribe | OpenAI | $0.0340 | 3.4 | $2.04 per hour of audio |
Voices
Voices are charged per character the agent speaks. The per-minute price assumes about 500 characters a minute.
| Voice | Provider | Per minute | Credits / min | Unit price |
|---|---|---|---|---|
| S2 Pro | Fish Audio | $0.0150 | 1.5 | $0.030 per 1,000 characters |
| GPT-4o mini TTS | OpenAI | $0.0150 | 1.5 | $0.030 per 1,000 characters |
| AssemblyAI-TTS-1 | AssemblyAI | $0.0150 | 1.5 | $0.030 per 1,000 characters |
| Flash v2.5 | ElevenLabs | $0.1500 | 15 | $0.300 per 1,000 characters |
| Turbo v2.5 | ElevenLabs | $0.1500 | 15 | $0.300 per 1,000 characters |
| v4 Turbo | ElevenLabs | $0.3000 | 30 | $0.600 per 1,000 characters |
| Multilingual v2 | ElevenLabs | $0.3000 | 30 | $0.600 per 1,000 characters |
Phone lines
Calls are priced by the country you dial, at the carrier's rate for that country. The index lists all 204 countries Callable can call; here are the most common:
| Calling | Dial code | Per minute | Credits / min |
|---|---|---|---|
| United States | +1 | $0.0140 | 1.4 |
| Canada | +1 | $0.0140 | 1.4 |
| United Kingdom | +44 | $0.0442 | 4.4 |
| Australia | +61 | $0.0702 | 7.0 |
| Germany | +49 | $0.0842 | 8.4 |
| France | +33 | $0.1068 | 11 |
| India | +91 | $0.0742 | 7.4 |
| Pakistan | +92 | $0.1680 | 17 |
| Bangladesh | +880 | $0.1990 | 20 |
| Saudi Arabia | +966 | $0.4560 | 46 |
| United Arab Emirates | +971 | $0.5040 | 50 |
| Nigeria | +234 | $0.5742 | 57 |
For any other country, search the Phone lines list on models.callableio.com. A few countries the carrier doesn't publish a rate for (Kiribati, Nauru, Sao Tome and Principe, Tuvalu, Vatican City) are billed at a cautious standard rate of $1.68 a minute.
You can also buy a local number in any of these countries. Its monthly price comes live from the carrier when you search: their monthly price plus $2, at least $3 a month. See Phone numbers.
Images
Not part of a call: when a business's website has no logo, Catalyst makes the new agent's profile picture, one 1024×1024 image at low quality. It tries these models in order and uses the first one available. Priced per picture; free while Catalyst is in preview.
| Model | Provider | Per picture | Credits | Unit price |
|---|---|---|---|---|
| GPT Image 1 mini | OpenAI | $0.0048 | 0.48 | $4.00 in · $16.00 out per 1M tokens (272 per picture) |
| GPT Image 1 | OpenAI | $0.0228 | 2.3 | $10.00 in · $80.00 out per 1M tokens (272 per picture) |
| DALL·E 3 | OpenAI | $0.0800 | 8.0 | $0.080 per 1024×1024 picture |
Response time
For language models the index also shows a response time: the median time from the caller's last word to the agent's first sound, on real Callable phone calls over the last 30 days. It includes the transcriber and the voice, so it's what a caller actually waits.
A model shows a number once 25 real calls have used it. Until then it says Measuring.
When prices change
Providers change their prices from time to time, and Callable checks them regularly. The date at the top of the index shows the last check. A new price applies to calls made after it, never to calls already made.
Questions
Why does my Usage page show a different cost per minute? The index prices a typical minute. Your agent's instructions, Knowledge Base, how much it talks and how long calls last all move the real number. Usage shows exactly what each call cost.
Which part should I change to save money? Usually the voice or the language model. Switching voice or model is a dropdown in the agent builder: see Voice and Agents.
Are these the prices I pay in credits? Yes. One credit is one cent, so $0.01 a minute is 1 credit a minute. See Pricing & billing.
Next steps
- models.callableio.com: the full index and calculator
- Pricing & billing: packs, credits and top-ups
- Voice and Transcriber: choosing each part
- Phone numbers: buying a number for your agent