---
id: amazon-bedrock
title: Amazon Bedrock
sidebar_label: Amazon Bedrock
---

`deepeval` supports Amazon Bedrock models that are available through the Bedrock Runtime Converse API for all evaluation metrics. To get started, you'll need to set up your AWS credentials.

:::note
`AmazonBedrockModel` requires `aiobotocore` and `botocore`. `deepeval` will prompt you to install them if they are missing.
:::

### Setting Up Your API Key

To use Amazon Bedrock for `deepeval`'s LLM-based evaluations (metrics evaluated using an LLM), provide your `AWS_ACCESS_KEY_ID` and `AWS_SECRET_ACCESS_KEY` in the CLI:

```bash
export AWS_ACCESS_KEY_ID=<your-aws-access-key-id>
export AWS_SECRET_ACCESS_KEY=<your-aws-secret-access-key>

```

Alternatively, if you're working in a notebook environment (e.g., Jupyter or Colab), set your keys in a cell:

```bash
%env AWS_ACCESS_KEY_ID=<your-aws-access-key-id>
%env AWS_SECRET_ACCESS_KEY=<your-aws-secret-access-key>
```

### Python

To use Amazon bedrock models for `deepeval` metrics, define an `AmazonBedrockModel` and specify the model you want to use.

<Tabs items={["Python", "ENV"]}>
<Tab value="Python">

```python
from deepeval.models import AmazonBedrockModel
from deepeval.metrics import AnswerRelevancyMetric

model = AmazonBedrockModel(
    model="anthropic.claude-3-opus-20240229-v1:0",
    region="us-east-1",
    generation_kwargs={"temperature": 0},
)
answer_relevancy = AnswerRelevancyMetric(model=model)
```

</Tab>
<Tab value="ENV">

To use any Amazon Bedrock model directly in `deepeval`, set the `USE_AWS_BEDROCK_MODEL=1` in your `env` and simply pass the name of your desired model in your metric initialization:

```python
from deepeval.metrics import AnswerRelevancyMetric

answer_relevancy = AnswerRelevancyMetric(
    model="anthropic.claude-3-opus-20240229-v1:0",
)
```

You should also set the other necessary vars like `AWS_ACCESS_KEY_ID`, `AWS_SESSION_TOKEN`, ..etc. to be able to use the Amazon Bedrock models as shown above.

</Tab>
</Tabs>

There are **ZERO** mandatory and **SEVEN** optional parameters when creating an `AmazonBedrockModel`:

- [Optional] `model`: A string specifying the bedrock model identifier to call (e.g. `anthropic.claude-3-opus-20240229-v1:0`). Defaults to `AWS_BEDROCK_MODEL_NAME` if not passed; raises an error at runtime if unset.
- [Optional] `region`: A string specifying the AWS region hosting your Bedrock endpoint (e.g. `us-east-1`). Defaults to `AWS_BEDROCK_REGION` if not passed; raises an error at runtime if unset.
- [Optional] `aws_access_key_id`: A string specifiying your AWS Access Key ID. Defaults to `AWS_ACCESS_KEY_ID` if not passed; if still omitted, falls back to the AWS default credentials chain.
- [Optional] `aws_secret_access_key`: A string specifiying your AWS Secret Access Key. Defaults to `AWS_SECRET_ACCESS_KEY` if not passed; if still omitted, falls back to the AWS default credentials chain.
- [Optional] `cost_per_input_token`: A float specifying the per-input-token cost in USD. Defaults to `AWS_BEDROCK_COST_PER_INPUT_TOKEN` if available in `deepeval`'s model cost registry, else `None`.
- [Optional] `cost_per_output_token`: A float specifying the per-output-token cost in USD. Defaults to `AWS_BEDROCK_COST_PER_OUTPUT_TOKEN` if available in `deepeval`'s model cost registry, else `None`.
- [Optional] `generation_kwargs`: A dictionary of generation parameters that will be sent to Bedrock as `inferenceConfig`. Available keys may vary by the Bedrock model you choose. See the [AWS Bedrock inference parameters docs](https://docs.aws.amazon.com/bedrock/latest/userguide/inference-parameters.html).

Parameters may be explicitly passed to the model at initialization time, or configured with optional settings. The **mandatory** parameters are required at runtime, but you can provide them either explicitly as constructor arguments, **or** via `deepeval` settings / environment variables (constructor args take precedence). See [Environment variables and settings](/docs/evaluation-flags-and-configs#model-settings-aws-amazon-bedrock) for the Bedrock-related environment variables.

:::tip
Pass generation parameters like `temperature`, `topP`, or `maxTokens` via `generation_kwargs` (they are sent as `inferenceConfig`).

Extra `**kwargs` passed to `AmazonBedrockModel(...)` are forwarded to the underlying Bedrock client (aiobotocore/botocore) and are **not** treated as generation parameters.
:::

### Available Amazon Bedrock Models

:::note
This list only displays some of the available models. For a comprehensive list, refer to the Amazon Bedrock's official documentation.
:::

Below is a list of commonly used Amazon Bedrock foundation models:

- `anthropic.claude-3-opus-20240229-v1:0`
- `anthropic.claude-3-sonnet-20240229-v1:0`
- `anthropic.claude-opus-4-20250514-v1:0`
- `anthropic.claude-opus-4-1-20250805-v1:0`
- `anthropic.claude-sonnet-4-20250514-v1:0`
- `anthropic.claude-sonnet-4-5-20250929-v1:0`
- `anthropic.claude-haiku-4-5-20251001-v1:0`
- `amazon.titan-text-express-v1`
- `amazon.titan-text-premier-v1:0`
- `amazon.nova-micro-v1:0`
- `amazon.nova-lite-v1:0`
- `amazon.nova-pro-v1:0`
- `amazon.nova-premier-v1:0`
- `meta.llama4-maverick-17b-instruct-v1:0`
- `meta.llama4-maverick-17b-instruct-128k-v1:0`
- `meta.llama4-scout-17b-instruct-v1:0`
- `meta.llama4-scout-17b-instruct-128k-v1:0`
- `mistral.mistral-large-2407-v1:0`
- `mistral.mistral-large-2411-v1:0`
- `mistral.pixtral-large-2411-v1:0`
- `mistral.pixtral-large-2502-v1:0`
- `mistral.pixtral-large-2511-v1:0`
- `openai.gpt-oss-20b-1:0`
- `openai.gpt-oss-120b-1:0`