Free access to all the language models with request rate limitation and regular captcha completion
At most 200 generated tokens per request
Speech to text transcription duration is limited to 3 minutes
All requests are logged on our servers
Standard Plan
Access to all the language models without rate limitation
No limitation on the number of generated tokens
No log of the request content is kept on our servers
Credit based hence no unexpected expense
Simple pricing based on the number of input and generated
tokens
Unit:
Large Language Models:
Model
Price per million input tokens
Price per million generated tokens
MADLAD400 7B
0.20 credits
2.00 credits
GPT-J 6B
0.20 credits
2.00 credits
Mistral 7B
0.20 credits
2.00 credits
Llama3 8B
0.20 credits
2.00 credits
Llama3 8B instruct
0.20 credits
2.00 credits
Mixtral 47B instruct
0.40 credits
4.00 credits
Llama2 70B
0.70 credits
7.00 credits
Image Generation:
Model
Price per generated image
Stable Diffusion
0.005000 credits
Speech to Text Transcription:
Model
Price per minute
Whisper Large V3
0.003000 credits
Large Language Models:
Model
Price per million input tokens
Price per million generated tokens
MADLAD400 7B
$ 0.20
$ 2.00
GPT-J 6B
$ 0.20
$ 2.00
Mistral 7B
$ 0.20
$ 2.00
Llama3 8B
$ 0.20
$ 2.00
Llama3 8B instruct
$ 0.20
$ 2.00
Mixtral 47B instruct
$ 0.40
$ 4.00
Llama2 70B
$ 0.70
$ 7.00
Image Generation:
Model
Price per generated image
Stable Diffusion
$ 0.005000
Speech to Text Transcription:
Model
Price per minute
Whisper Large V3
$ 0.003000
Large Language Models:
Model
Price per million input tokens
Price per million generated tokens
MADLAD400 7B
€ 0.17
€ 1.71
GPT-J 6B
€ 0.17
€ 1.71
Mistral 7B
€ 0.17
€ 1.71
Llama3 8B
€ 0.17
€ 1.71
Llama3 8B instruct
€ 0.17
€ 1.71
Mixtral 47B instruct
€ 0.34
€ 3.42
Llama2 70B
€ 0.60
€ 5.98
Image Generation:
Model
Price per generated image
Stable Diffusion
€ 0.004270
Speech to Text Transcription:
Model
Price per minute
Whisper Large V3
€ 0.002560
Pricing FAQ
How does the pricing work ?
The price of a request is determined from the number of input
tokens (i.e. the ones you send) multiplied by the input token price
summed with the number of generated tokens multiplied by the
generated token price.
Generated tokens are more expensive than input tokens because
they require more computation resources.
A token is the unit of information processed by a language
model. It depends on the model and corresponds to about 4
characters for English texts.
For image generation, the price is determinated by the number of generated images.
For speech to text transcription, the price is determinated by the duration of the input audio rounded to the next second.
How is the payment handled ?
TextSynth uses a credit system with
pre-purchased credits. This way you never pay more than
expected. The payment is handled
by Stripe. The minimum purchase
is of €20 / $20.
Can I pay in euros ?
Yes, TextSynth accepts payments in euros
too.
How long last the credits ?
Purchased credits last one year. If they are
not fully used after this duration they are no longer valid.
What happens when my number of credits goes down to zero ?
Your plan is automatically set to the Free Plan
with the associated limitations. When the number of credit is
greater than zero the Standard Plan is enabled again.
My balance used to be displayed in € or $. It is now
displayed in credits. What happened ?
On June 30th, 2023, all account balances were
converted to credits. $1 converted to 1.00 credit, €1 converted to
1.17 credits.