Purchasing no-commitment provisioned throughput for custom Amazon Text Titan Express Model

Question

#### Context
I created a custom Amazon Text Titan Express Model via continued pretraining on amazon bedrock and I would like to evaluate it. 
I have been trying to purchase provisioned throughput for my model since I found that such models have to be deployed via provisioned throughput for running inference.
I tried to purchase using "No commitment" to pay hourly since I'd like to test and iterate on the model before provisioning it for a longer period.

#### Concern
I am unable to purchase using "No commitment" and I see an error stating that the Amazon Titan Text Express model is not supported in "No commitment provision throughput".

I am looking for advice on deploying my custom model with "No commitment". Please guide me on how to proceed. Please let me know in case "No commitment" is not permitted, in which case I will try other commitment periods.

re:Post Agent · Answer

Based on the information available, you should be able to purchase no-commitment Provisioned Throughput for your custom Amazon Titan Text Express model. Multiple sources confirm that the "No commitment" option is available for custom models in Amazon Bedrock, including fine-tuned models like yours.

Here's how to purchase no-commitment Provisioned Throughput for your custom model:

1. On the Amazon Bedrock console, navigate to "Custom models" in the navigation pane
2. Select your custom model and choose "Purchase provisioned throughput"
3. Enter a name for your Provisioned Throughput
4. Ensure the correct custom model is selected
5. Under "Commitment term & model units", select "No commitment" and set 1 model unit (or the amount you need)
6. Review the estimated cost and confirm the purchase

Alternatively, you can use the AWS CLI with this command:
```
aws bedrock create-provisioned-model-throughput \
--model-id [Your-Custom-Model-ARN] \
--provisioned-model-name [Your-Provisioned-Model-Name] \
--model-units 1
```

After purchasing, allow some time for the Provisioned Throughput to be created. You can check its status using:
```
aws bedrock get-provisioned-model-throughput \
--provisioned-model-id [provisioned-model-arn]
```

When the status shows "InService", you can run inference with your custom model.

The error you're encountering is unexpected since the documentation specifically mentions that a flexible "No commitment" option is available for custom models. If you continue to experience this issue, you might want to contact AWS Support to help troubleshoot the specific error you're encountering.  
**Sources**  
[Code samples for model customization - Amazon Bedrock](https://docs.aws.amazon.com/bedrock/latest/userguide/model-customization-code-samples.html)  
[Fine-tune and deploy language models with Amazon SageMaker Canvas and Amazon Bedrock | AWS Machine Learning Blog](https://aws.amazon.com/blogs/machine-learning/fine-tune-and-deploy-language-models-with-amazon-sagemaker-canvas-and-amazon-bedrock/)  
[Fine-tune LLMs with synthetic data for context-based Q&A using Amazon Bedrock | AWS Machine Learning Blog](https://aws.amazon.com/blogs/machine-learning/fine-tune-llms-with-synthetic-data-for-context-based-qa-using-amazon-bedrock/)  
[Security best practices to consider while fine-tuning models in Amazon Bedrock | AWS Machine Learning Blog](https://aws.amazon.com/blogs/machine-learning/security-best-practices-to-consider-while-fine-tuning-models-in-amazon-bedrock/)

Purchasing no-commitment provisioned throughput for custom Amazon Text Titan Express Model

Context

Concern

Add your answer

Relevant content