Skip to main content

IBM watsonx.ai

LiteLLM supports all IBM watsonx.ai foundational models and embeddings.

Environment Variables​

os.environ["WATSONX_URL"] = ""  # (required) Base URL of your WatsonX instance
# (required) either one of the following:
os.environ["WATSONX_APIKEY"] = "" # IBM cloud API key
os.environ["WATSONX_TOKEN"] = "" # IAM auth token
# optional - can also be passed as params to completion() or embedding()
os.environ["WATSONX_PROJECT_ID"] = "" # Project ID of your WatsonX instance
os.environ["WATSONX_DEPLOYMENT_SPACE_ID"] = "" # ID of your deployment space to use deployed models
os.environ["WATSONX_ZENAPIKEY"] = "" # Zen API key (use for long-term api token)

See here for more information on how to get an access token to authenticate to watsonx.ai.

Usage​

Open In Colab 
import os
from litellm import completion

os.environ["WATSONX_URL"] = ""
os.environ["WATSONX_APIKEY"] = ""

## Call WATSONX `/text/chat` endpoint - supports function calling
response = completion(
  model="watsonx/meta-llama/llama-3-1-8b-instruct",
  messages=[{ "content": "what is your favorite colour?","role": "user"}],
  project_id="<my-project-id>" # or pass with os.environ["WATSONX_PROJECT_ID"]
)

## Call WATSONX `/text/generation` endpoint - not all models support /chat route. 
response = completion(
  model="watsonx/ibm/granite-13b-chat-v2",
  messages=[{ "content": "what is your favorite colour?","role": "user"}],
  project_id="<my-project-id>"
)

Usage - Streaming​

import os
from litellm import completion

os.environ["WATSONX_URL"] = ""
os.environ["WATSONX_APIKEY"] = ""
os.environ["WATSONX_PROJECT_ID"] = ""

response = completion(
  model="watsonx/meta-llama/llama-3-1-8b-instruct",
  messages=[{ "content": "what is your favorite colour?","role": "user"}],
  stream=True
)
for chunk in response:
  print(chunk)

Example Streaming Output Chunk​

{
  "choices": [
    {
      "finish_reason": null,
      "index": 0,
      "delta": {
        "content": "I don't have a favorite color, but I do like the color blue. What's your favorite color?"
      }
    }
  ],
  "created": null,
  "model": "watsonx/ibm/granite-13b-chat-v2",
  "usage": {
    "prompt_tokens": null,
    "completion_tokens": null,
    "total_tokens": null
  }
}

Usage - Models in deployment spaces​

Models that have been deployed to a deployment space (e.g.: tuned models) can be called using the deployment/<deployment_id> format (where <deployment_id> is the ID of the deployed model in your deployment space).

The ID of your deployment space must also be set in the environment variable WATSONX_DEPLOYMENT_SPACE_ID or passed to the function as space_id=<deployment_space_id>.

import litellm
response = litellm.completion(
    model="watsonx/deployment/<deployment_id>",
    messages=[{"content": "Hello, how are you?", "role": "user"}],
    space_id="<deployment_space_id>"
)

Usage - Embeddings​

LiteLLM also supports making requests to IBM watsonx.ai embedding models. The credential needed for this is the same as for completion.

from litellm import embedding

response = embedding(
    model="watsonx/ibm/slate-30m-english-rtrvr",
    input=["What is the capital of France?"],
    project_id="<my-project-id>"
)
print(response)
# EmbeddingResponse(model='ibm/slate-30m-english-rtrvr', data=[{'object': 'embedding', 'index': 0, 'embedding': [-0.037463713, -0.02141933, -0.02851813, 0.015519324, ..., -0.0021367231, -0.01704561, -0.001425816, 0.0035238306]}], object='list', usage=Usage(prompt_tokens=8, total_tokens=8))

OpenAI Proxy Usage​

Here's how to call IBM watsonx.ai with the LiteLLM Proxy Server

1. Save keys in your environment​

export WATSONX_URL=""
export WATSONX_APIKEY=""
export WATSONX_PROJECT_ID=""

2. Start the proxy​

$ litellm --model watsonx/meta-llama/llama-3-8b-instruct

# Server running on http://0.0.0.0:4000

3. Test it​

curl --location 'http://0.0.0.0:4000/chat/completions' \
--header 'Content-Type: application/json' \
--data ' {
      "model": "llama-3-8b",
      "messages": [
        {
          "role": "user",
          "content": "what is your favorite colour?"
        }
      ]
    }
'

Authentication​

Passing credentials as parameters​

You can also pass the credentials as parameters to the completion and embedding functions.

import os
from litellm import completion

response = completion(
            model="watsonx/ibm/granite-13b-chat-v2",
            messages=[{ "content": "What is your favorite color?","role": "user"}],
            url="",
            api_key="",
            project_id=""
)

Using Zen API Key​

You can use a Zen API key for long-term authentication instead of generating IAM tokens. Pass it either as an environment variable or as a parameter:

import os
from litellm import completion

# Option 1: Set as environment variable
os.environ["WATSONX_ZENAPIKEY"] = "your-zen-api-key"

response = completion(
    model="watsonx/ibm/granite-13b-chat-v2",
    messages=[{"content": "What is your favorite color?", "role": "user"}],
    project_id="your-project-id"
)

# Option 2: Pass as parameter
response = completion(
    model="watsonx/ibm/granite-13b-chat-v2",
    messages=[{"content": "What is your favorite color?", "role": "user"}],
    zen_api_key="your-zen-api-key",
    project_id="your-project-id"
)

Using with LiteLLM Proxy via OpenAI client:

import openai

client = openai.OpenAI(
    api_key="sk-1234",  # LiteLLM proxy key
    base_url="http://0.0.0.0:4000"
)

response = client.chat.completions.create(
    model="watsonx/ibm/granite-3-3-8b-instruct",
    messages=[{"role": "user", "content": "What is your favorite color?"}],
    max_tokens=2048,
    extra_body={
        "project_id": "your-project-id",
        "zen_api_key": "your-zen-api-key"
    }
)

See IBM documentation for more information on generating Zen API keys.

Supported IBM watsonx.ai Models​

Here are some examples of models available in IBM watsonx.ai that you can use with LiteLLM:

Mode NameCommand
Flan T5 XXLcompletion(model=watsonx/google/flan-t5-xxl, messages=messages)
Flan Ul2completion(model=watsonx/google/flan-ul2, messages=messages)
Mt0 XXLcompletion(model=watsonx/bigscience/mt0-xxl, messages=messages)
Gpt Neoxcompletion(model=watsonx/eleutherai/gpt-neox-20b, messages=messages)
Mpt 7B Instruct2completion(model=watsonx/ibm/mpt-7b-instruct2, messages=messages)
Starcodercompletion(model=watsonx/bigcode/starcoder, messages=messages)
Llama 2 70B Chatcompletion(model=watsonx/meta-llama/llama-2-70b-chat, messages=messages)
Llama 2 13B Chatcompletion(model=watsonx/meta-llama/llama-2-13b-chat, messages=messages)
Granite 13B Instructcompletion(model=watsonx/ibm/granite-13b-instruct-v1, messages=messages)
Granite 13B Chatcompletion(model=watsonx/ibm/granite-13b-chat-v1, messages=messages)
Flan T5 XLcompletion(model=watsonx/google/flan-t5-xl, messages=messages)
Granite 13B Chat V2completion(model=watsonx/ibm/granite-13b-chat-v2, messages=messages)
Granite 13B Instruct V2completion(model=watsonx/ibm/granite-13b-instruct-v2, messages=messages)
Elyza Japanese Llama 2 7B Instructcompletion(model=watsonx/elyza/elyza-japanese-llama-2-7b-instruct, messages=messages)
Mixtral 8X7B Instruct V01 Qcompletion(model=watsonx/ibm-mistralai/mixtral-8x7b-instruct-v01-q, messages=messages)

For a list of all available models in watsonx.ai, see here.

Supported IBM watsonx.ai Embedding Models​

Model NameFunction Call
Slate 30membedding(model="watsonx/ibm/slate-30m-english-rtrvr", input=input)
Slate 125membedding(model="watsonx/ibm/slate-125m-english-rtrvr", input=input)

For a list of all available embedding models in watsonx.ai, see here.