Oracle Cloud Infrastructure Generative AI

https://docs.oracle.com/en-us/iaas/Content/generative-ai/
Admin
Oracle Cloud Infrastructure Generative AI is a fully managed service providing state-of-the-art,...
Tokens:355,594
Snippets:2,098
Trust Score:9.7
Update:1 day ago
Show doc for...
Tokens:
Raw
### Create and activate virtual environment and install OpenAI SDK

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/routing-profile-tutorial.htm

Creates a Python 3.11 virtual environment, activates it, and installs the OpenAI Python SDK. If .venv already exists, activate it and run the installation command. The OCI Python SDK is not needed for these examples.

```bash
uv venv --python 3.11 .venv
source .venv/bin/activate
uv pip install --upgrade openai
```

--------------------------------

### Example output of pip list

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/setup-oci-api-auth.htm

Example output showing oci-cli with a version placeholder.

```text
oci-cli                  <version>
```

--------------------------------

### Install OCI CLI

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/setup-oci-api-auth.htm

Install the OCI CLI and Python SDK using pip.

```bash
pip install oci-cli
```

--------------------------------

### Create Your First Conversation (OpenAI SDK Example)

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/conversations-api.htm

Example demonstrating how to create a conversation and send responses using the OpenAI SDK with the OCI OpenAI-compatible endpoint. This example shows the usage of client.conversations.create and client.responses.create.

```APIDOC
## Create Your First Conversation

### Description
This example demonstrates how to create a conversation and interact with the OCI Responses API using the OpenAI SDK with the OCI OpenAI-compatible endpoint.

### Prerequisites
- An OCI Generative AI API key
- A project OCID

### Configuration
- **base_url**: `https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/openai/v1`
- **api_key**: Your OCI Generative AI API key
- **project**: Your OCI Generative AI project OCID

### Python Example
```python
from openai import OpenAI

client = OpenAI(
    base_url="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/openai/v1",
    api_key="<your-api-key>", # replace with your Generative AI API Key
    project="ocid1.generativeaiproject.oc1.us-chicago-1.xxxxxxxx",
)

# create a conversation
conversation = client.conversations.create(
    metadata={"topic": "offsite"}
)
print("Conversation ID:", conversation.id)

# first question
response1 = client.responses.create(
    model="openai.gpt-oss-120b",
    input="Give me three ideas for a team offsite.",
    conversation=conversation.id,
)
print("Response 1:", response1.output_text)

# second question
response2 = client.responses.create(
    model="openai.gpt-oss-120b",
    input="Make the second idea more budget friendly.",
    conversation=conversation.id,
)
print("Response 2:", response2.output_text)
```

### Notes
- `base_url` points to the OCI OpenAI-compatible endpoint.
- `client.responses.create(...)` calls the OCI Responses API.
- `project` identifies an already created OCI Generative AI project for the request.
```

--------------------------------

### Install OpenAI SDK

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/get-started-agents.htm

Install the official OpenAI SDK. Use the latest version to invoke the Responses API.

```bash
pip install openai
```

--------------------------------

### Verify OCI CLI configuration with 'oci os ns get'

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/setup-oci-api-auth.htm

Run 'oci os ns get' to get the namespace of your tenancy, verifying the configuration file works.

```bash
oci os ns get
```

--------------------------------

### Install the Langfuse SDK

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/get-started-agents.htm

Install the Langfuse SDK using pip. This is required for integrating the OCI Responses API with Langfuse for observability.

```bash
pip install langfuse
```

--------------------------------

### tool_choice Property JSON Example

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/get-started-agents.htm

Illustrates the tool_choice property to guide tool behavior. The example shows an allowed_tools type with mode auto and a list of two function tools: get_calendar_events and purchase_tickets.

```json
{
  "tool_choice": {
    "type": "allowed_tools",
    "mode": "auto",
    "tools": [
        { "type": "function", "name": "get_calendar_events" },
        { "type": "function", "name": "purchase_tickets" }
    ]
  }
}
```

--------------------------------

### Verify OCI CLI installation

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/setup-oci-api-auth.htm

Lists installed packages to confirm oci-cli is installed.

```bash
pip list
```

--------------------------------

### Example OCI config file

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/setup-oci-api-auth.htm

Example configuration file for the DEFAULT profile, containing user OCID, fingerprint, tenancy OCID, region, and key file path.

```ini
[DEFAULT]
user=<ocid1.user.oc1..xxx>
fingerprint=<xx:xx:xx:xx>
tenancy=<ocid1.tenancy.oc1..xxx>
region=us-chicago-1
key_file=<path-to-private-key.pem>
```

--------------------------------

### Python example: Use X Search tool with OCI Responses API

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/get-started-agents.htm

Shows how to call the OCI Responses API with the x_search tool to search X posts. Configure the OpenAI client with OCI authentication and specify tool parameters like allowed_x_handles, from_date, to_date, and enable_image_understanding.

```python
from openai import OpenAI
from oci_genai_auth import OciUserPrincipalAuth
import httpx

client = OpenAI(
    base_url="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/openai/v1",
    api_key="not-used",
    project="ocid1.generativeaiproject.oc1.us-chicago-1.xxxxxxxx",
    http_client=httpx.Client(
        auth=OciUserPrincipalAuth(
            config_file="~/.oci/config",
            profile_name="DEFAULT"
        )
    )
)

response = client.responses.create(
    model="xai.grok-4.3",
    tools=[{
            "type": "x_search",
            "enable_image_understanding": True,
            "allowed_x_handles": ["example-person1-handle", "example-person2-handle"],
            "from_date": "2026-01-01",
            "to_date": "2026-04-30",
            "enable_video_understanding": True,
        }],
    input="Give me a summary of the last 10 posts by @example-person1."
)

print(response.output_text)
```

--------------------------------

### Install langchain package

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/setup-oci-langchain.htm

Installs or upgrades the `langchain` package to the latest version using pip.

```bash
pip install -U langchain
```

--------------------------------

### Project Directory Structure

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/prepare-artifacts.htm

Shows the recommended layout for the agent project directory, including main.py, requirements.txt, Dockerfile, .dockerignore, and __init__.py.

```text
project_directory/
├── main.py # Your main agent code
├── requirements.txt # Dependencies for your agent
├── Dockerfile # Dockerfile for building the image
├── .dockerignore # Files not included in the image
└── __init__.py # Makes the directory a Python package

```

--------------------------------

### OciSessionAuth Example

Source: https://docs.oracle.com/en-us/iaas/Content/generative-ai/get-started-agents.htm

Uses session-based authentication for local development. Requires the oci_genai_auth package and an OCI session profile.

```python
from openai import OpenAI
from oci_genai_auth import OciSessionAuth
import httpx

client = OpenAI(
    base_url="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/openai/v1",  # update region if needed
    api_key="not-used",
    project="ocid1.generativeaiproject.oc1.us-chicago-1.xxxxxxxx",  # project OCID created earlier
    http_client=httpx.Client(auth=OciSessionAuth(profile_name="DEFAULT"))  # update profile if needed
)

response = client.responses.create(
    model="xai.grok-4.3",
    input="Write a one-sentence explanation of what a database is."
)

print(response.output_text)
```
Tokens:
Raw