Self-service pricing FAQ
You need to buy a pack to add money to your balance in GPT for Work add-ons. Your balance can be used with all models: models that do not require an API key, and models that require an API key.
OpenAI's ChatGPT Plus subscription is neither necessary nor sufficient for using the GPT for Work add-ons.
How does the balance work?
Your balance will decrease according to your token consumption, which includes both input and output tokens. Your balance can be used with any model you want.
In GPT for Sheets, you can turn spending off on all your spreadsheets to freeze your balance. This helps you avoid unexpected costs caused by Sheets auto-refreshes.
What is a token?
A token is a small piece of text that can represent a word or subword, a punctuation sign or a symbol. Models split all input and output into tokens for efficient processing. The token is the basic unit of usage of model providers.
The cost of a token depends on the model used.
What does cost depend on?
Cost depends on the model used and the number of input and output tokens sent to and from the model provider (except for the GPT_WEB function and the Search the web bulk tool, which are priced per 1000 executions).
Cost = all input costs + all output costs, where:
- Input = Text and images sent to the model, including custom instructions and prompt templates that GPT for Work uses under the hood (which are very small)
- Output = Results returned by the model
Use the cost estimator to estimate what your use case will cost.
Which models can I use?
Check our full and up-to-date list of supported models. You can switch models at any time.
OpenAI Assistants, fine-tuned models, and some models are only available if you set an API key.
Do I need to buy a pack even though I'm using my own OpenAI API key?
Yes, you need to buy a pack. When you use a model that requires an OpenAI API key, you have two costs:
- The GPT for Work add-on cost paid to Talarian as per our pricing table. This cost is deducted from your balance.
- The OpenAI API cost paid directly to OpenAI (see this page for up-to-date pricing information).
Your total cost is the sum of those two costs, which is cheaper than using the same model without an API key.
You can use the cost estimator to estimate what your use case will cost.
Packs bought in GPT for Excel and GPT for Word can be used with both add-ins. Packs bought in GPT for Sheets can only be used with GPT for Sheets.
When should I use my own API key?
If you are familiar with API keys and if you have an API key with high enough rate limits, it is advised to set your own API key in order to:
- Save on total cost on equivalent models compared to not using an API key
- Get access to more models
- Benefit from your own rate limits
- Use your own OpenAI Assistants
- Use your own fine-tuned models on GPT for Sheets
Token count and language
Token counts vary by language because the way text is split into tokens differs from one language to another.
On average (figures from this study):
- English: 1 word ≈ 1.3 tokens
- French: 1 word ≈ 2 tokens
- German: 1 word ≈ 2.1 tokens
- Spanish: 1 word ≈ 2.1 tokens
- Chinese: 1 word ≈ 2.5 tokens
- Russian: 1 word ≈ 3.3 tokens
- Vietnamese: 1 word ≈ 3.3 tokens
- Arabic: 1 word ≈ 4 tokens
- Hindi: 1 word ≈ 6.4 tokens
Refer to the study for other languages.
These figures are presented for estimation purposes only and are not guaranteed.
Can I set a spending limit?
It is not possible to set an arbitrary spending limit. However, since you cannot spend more than your current balance, your balance always acts as a spending limit.
Can I buy for my team?
You can add members to your space from the dashboard. Members of a space all share the same balance, which you get from buying packs for the space.
If you need help, contact support.
What is the cost of GPT_WEB and the Search the web tools?
GPT_WEB and the Search the web bulk tool are priced per 1000 executions instead of tokens. In this case, input and output sizes do not matter.
The price is $15 or €15 per 1000 executions.
Are errors billed?
You will not be charged (neither by us nor by the model provider) for executions where the model provider returns an error.
Are timeouts billed?
When using GPT functions in Google Sheets, a timeout can occur if the model provider's output takes longer than 30 seconds to return, causing Google Sheets to block the response.
When a timeout happens, this is how you are charged:
- If you are using an API key, you will be charged by the model provider but not by us.
- If you're not using an API key, you will be charged by us for the input only and we will cover the cost of the output since we are not able to send it back to you.
You can enable GPT for Sheets Safe mode to avoid timeouts.
Why do I see my balance quickly go down and then go back up?
When you launch an execution with GPT for Work add-ons, the output size and thus full cost is still unknown. Your balance is pre-debited to make sure it can cover the full cost. If the full cost is lower than what was pre-debited, the remainder will be immediately re-credited. If you execute many GPT functions at once, or run bulk tools on a large number of rows, you may see your balance go down very quickly, and then go back up.
How can I avoid unexpected costs when spreadsheets auto-refresh in Google Sheets?
To prevent unexpected costs due to automatic recalculations in Google Sheets, refer to our guide on Avoiding Recalculations.
Who pays if I share my spreadsheet with others?
GPT for Excel users are charged for all the executions they perform.
GPT for Sheets users can share their spreadsheets with others, and be charged for the GPT function executions performed by others. See what role is required and who will be charged according to the spreadsheet location. Bulk tool executions are charged to the user who executes them, whether they are on a shared spreadsheet or not.
Is there a free plan?
No, there is no free plan. There is a free trial.
Can I get a refund?
We do not offer refunds. We encourage you to carefully consider how much balance you need before buying.