Pricing
Our Luminous models come in different sizes and prices. To find out which one suits your needs, you can refer to our Introduction section.
One token is roughly equivalent to a short word or a syllable. As a rule of thumb, 1000 tokens are about 720 words. Images always have 144 tokens, regardless of their size and resolution.
Credits are Aleph Alpha’s internal calculation unit. We use them to better allocate the costs of your usage and to provide you with a detailed overview of your spending.
All prices shown are in credits and the respective EUR equivalent.
Token-based Model Pricing
In the table below you can find the base prices per model.
| Model | Price per 1000 input tokens | Price per input image |
|---|---|---|
| Base | 0.03 (€0.006) | 0.03024 (€0.006048) |
| Extended | 0.045 (€0.009) | 0.04536 (€0,009072) |
| Supreme | 0.175 (€0.035) | - |
| Base-control | 0.0375 (€0.0075) | - |
| Extended-control | 0.05625 (€0.01125) | - |
| Supreme-control | 0.21875 (€0.04375) | - |
Task Pricing
Based on these base prices per model, we charge different rates for each task. We differentiate between input and output tokens. Input tokens are everything that you submit to our models through a prompt. Output tokens are all tokens that Luminous returns to you as a completion of your prompt. Input tokens are less costly for us to compute than output tokens. We try to minimize costs for our users as much as possible, which is why we only charge for the resources you actually use. There are no hidden fees for you. As not all tasks generate output tokens, multiplication factors for output tokens are displayed where applicable. The multiplicative factor for output tokens is applied on top of the task-specific factor for input tokens.
| Task | Factor input tokens | Additional factor output tokens |
|---|---|---|
| Complete | 1.0 | 1.1 |
| Evaluate | 1.1 | 1.1 |
| Embed | 1.3 | - |
| Semantic Embed | 1.3 | - |
| Q&A | 1.3 | 1.1 |
| Summarize | 1.3 | 1.1 |
| (De-) Tokenize | 0.5 | - |
Request-based Service Pricing
For some endpoints we have a request-based pricing model. This means that you will pay per request and not token-based. Prices below are in credits.
| Task | Price |
|---|---|
| Explain | 0.1 credits per request |
For example, an "Evaluate" task for 1000 input tokens on Luminous-extended costs 0.045 (base rate) * 1.1 (input task specific factor) = 0.0495 credits. Additionally, 1000 evaluate output tokens would then cost another 0.0495 * 1.1 (output task specific factor) = 0.05445 credits. Thus, the total cost for this 2000 (1000 input + 1000 output) token request would be 0.0495 + 0.05445 = 0.10395 credits.