Models
Speech, language and voice model prices
The full Wixzel Voice price list behind the engines and the compose builder, in the unit each model is billed in. Generated from the price book the meter bills against, so it cannot drift from what you are charged.
Last updated
| Model | Stage | Family | Unit price | Per minute |
|---|---|---|---|---|
| openrouter/anthropic/claude-haiku-4.5 | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0014 per 1k tokens in $0.0070 per 1k tokens out | $0.0067 |
| openrouter/anthropic/claude-sonnet-4.5 | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0042 per 1k tokens in $0.0210 per 1k tokens out | $0.0202 |
| openrouter/deepseek/deepseek-chat-v3.1 | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0003 per 1k tokens in $0.0013 per 1k tokens out | $0.0015 |
| openrouter/google/gemini-2.5-flash | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0004 per 1k tokens in $0.0035 per 1k tokens out | $0.0027 |
| openrouter/google/gemini-2.5-flash-lite | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0001 per 1k tokens in $0.0006 per 1k tokens out | $0.0006 |
| openrouter/gpt-4o-mini | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0002 per 1k tokens in $0.0008 per 1k tokens out | $0.0009 |
| openrouter/meta-llama/llama-3.3-70b-instruct | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0001 per 1k tokens in $0.0004 per 1k tokens out | $0.0006 |
| openrouter/mistralai/mistral-small-3.2-24b-instruct | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0001 per 1k tokens in $0.0003 per 1k tokens out | $0.0004 |
| openrouter/openai/gpt-4.1-mini | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0006 per 1k tokens in $0.0022 per 1k tokens out | $0.0024 |
| openrouter/openai/gpt-4o | Language model | Deepgram · OpenRouter · ElevenLabs | $0.0035 per 1k tokens in $0.0140 per 1k tokens out | $0.0151 |
| sarvam/sarvam-105b | Language model | Sarvam | $0.0004 per 1k tokens in $0.0011 per 1k tokens out | $0.0016 |
| deepgram/agent | Realtime model | realtime | $0.0018 per second | $0.1050 |
| google/gemini-live | Realtime model | realtime | $0.0042 per 1k audio tokens in $0.0168 per 1k audio tokens out $0.0007 per 1k tokens in $0.0028 per 1k tokens out | $0.0345 |
| deepgram/nova-3 | Speech-to-text | Deepgram · OpenRouter · ElevenLabs | $0.0002 per second | $0.0108 |
| deepgram/nova-3-multilingual | Speech-to-text | Deepgram · OpenRouter · ElevenLabs | $0.0002 per second | $0.0129 |
| sarvam/saaras-v3-realtime | Speech-to-text | Sarvam | $0.0001 per second | $0.0076 |
| elevenlabs/eleven_flash_v2_5 | Text-to-speech | Deepgram · OpenRouter · ElevenLabs | $0.0700 per 1k characters | $0.0630 |
| elevenlabs/eleven_multilingual_v2 | Text-to-speech | Deepgram · OpenRouter · ElevenLabs | $0.1400 per 1k characters | $0.1260 |
| elevenlabs/eleven_turbo_v2_5 | Text-to-speech | Deepgram · OpenRouter · ElevenLabs | $0.0700 per 1k characters | $0.0630 |
| sarvam/bulbul-v3 | Text-to-speech | Sarvam | $0.0457 per 1k characters | $0.0411 |
Add $0.012 per connected minute of orchestration to any composition. Telephony is your own SIP trunk and adds nothing here. Try combinations in the compose builder, or read how composing works.
Which models can be combined in one Wixzel Voice engine?
The speech-to-text provider selects the pipeline, and a pipeline runs one family end to end, so the three stages of a composed engine must come from the same family. classic is Deepgram · OpenRouter · ElevenLabs. sarvam is Sarvam. Realtime models are their own engine and are not composed.
How much does each model cost per minute?
- How are model prices on this page calculated?
- Each model is billed in its own unit: seconds of audio for speech-to-text, tokens in and out for language models, characters for text-to-speech, and audio or text tokens for realtime models. Tokens and characters are quoted per thousand here, as every provider quotes them; the meter itself works in millionths of a dollar per single unit, so nothing is rounded when you are charged. The per-minute column multiplies the unit price by the units a typical minute of conversation uses (15 characters per second spoken, 40 tokens in and 8 out per second, 25 audio tokens per second each way). The Wixzel Voice meter charges for the units that actually ran, so a real minute may cost less or more.
- How much does openrouter/anthropic/claude-haiku-4.5 cost per minute on Wixzel Voice?
- openrouter/anthropic/claude-haiku-4.5 (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0014 per 1k tokens in and $0.0070 per 1k tokens out, which is about $0.0067 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/anthropic/claude-sonnet-4.5 cost per minute on Wixzel Voice?
- openrouter/anthropic/claude-sonnet-4.5 (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0042 per 1k tokens in and $0.0210 per 1k tokens out, which is about $0.0202 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/deepseek/deepseek-chat-v3.1 cost per minute on Wixzel Voice?
- openrouter/deepseek/deepseek-chat-v3.1 (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0003 per 1k tokens in and $0.0013 per 1k tokens out, which is about $0.0015 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/google/gemini-2.5-flash cost per minute on Wixzel Voice?
- openrouter/google/gemini-2.5-flash (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0004 per 1k tokens in and $0.0035 per 1k tokens out, which is about $0.0027 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/google/gemini-2.5-flash-lite cost per minute on Wixzel Voice?
- openrouter/google/gemini-2.5-flash-lite (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0001 per 1k tokens in and $0.0006 per 1k tokens out, which is about $0.0006 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/gpt-4o-mini cost per minute on Wixzel Voice?
- openrouter/gpt-4o-mini (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0002 per 1k tokens in and $0.0008 per 1k tokens out, which is about $0.0009 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/meta-llama/llama-3.3-70b-instruct cost per minute on Wixzel Voice?
- openrouter/meta-llama/llama-3.3-70b-instruct (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0001 per 1k tokens in and $0.0004 per 1k tokens out, which is about $0.0006 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/mistralai/mistral-small-3.2-24b-instruct cost per minute on Wixzel Voice?
- openrouter/mistralai/mistral-small-3.2-24b-instruct (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0001 per 1k tokens in and $0.0003 per 1k tokens out, which is about $0.0004 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/openai/gpt-4.1-mini cost per minute on Wixzel Voice?
- openrouter/openai/gpt-4.1-mini (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0006 per 1k tokens in and $0.0022 per 1k tokens out, which is about $0.0024 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does openrouter/openai/gpt-4o cost per minute on Wixzel Voice?
- openrouter/openai/gpt-4o (language model, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0035 per 1k tokens in and $0.0140 per 1k tokens out, which is about $0.0151 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does sarvam/sarvam-105b cost per minute on Wixzel Voice?
- sarvam/sarvam-105b (language model, Sarvam family) is billed $0.0004 per 1k tokens in and $0.0011 per 1k tokens out, which is about $0.0016 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does deepgram/agent cost per minute on Wixzel Voice?
- deepgram/agent (realtime model, realtime family) is billed $0.0018 per second, which is about $0.1050 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does google/gemini-live cost per minute on Wixzel Voice?
- google/gemini-live (realtime model, realtime family) is billed $0.0042 per 1k audio tokens in and $0.0168 per 1k audio tokens out and $0.0007 per 1k tokens in and $0.0028 per 1k tokens out, which is about $0.0345 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does deepgram/nova-3 cost per minute on Wixzel Voice?
- deepgram/nova-3 (speech-to-text, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0002 per second, which is about $0.0108 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does deepgram/nova-3-multilingual cost per minute on Wixzel Voice?
- deepgram/nova-3-multilingual (speech-to-text, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0002 per second, which is about $0.0129 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does sarvam/saaras-v3-realtime cost per minute on Wixzel Voice?
- sarvam/saaras-v3-realtime (speech-to-text, Sarvam family) is billed $0.0001 per second, which is about $0.0076 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does elevenlabs/eleven_flash_v2_5 cost per minute on Wixzel Voice?
- elevenlabs/eleven_flash_v2_5 (text-to-speech, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0700 per 1k characters, which is about $0.0630 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does elevenlabs/eleven_multilingual_v2 cost per minute on Wixzel Voice?
- elevenlabs/eleven_multilingual_v2 (text-to-speech, Deepgram · OpenRouter · ElevenLabs family) is billed $0.1400 per 1k characters, which is about $0.1260 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does elevenlabs/eleven_turbo_v2_5 cost per minute on Wixzel Voice?
- elevenlabs/eleven_turbo_v2_5 (text-to-speech, Deepgram · OpenRouter · ElevenLabs family) is billed $0.0700 per 1k characters, which is about $0.0630 per minute at typical speech rates, before the $0.012 per minute orchestration fee.
- How much does sarvam/bulbul-v3 cost per minute on Wixzel Voice?
- sarvam/bulbul-v3 (text-to-speech, Sarvam family) is billed $0.0457 per 1k characters, which is about $0.0411 per minute at typical speech rates, before the $0.012 per minute orchestration fee.