IBM

import Icon from "@site/src/components/icon"; import PartialParams from '@site/docs/_partial-hidden-params.mdx'; import PartialConditionalParams from '@site/docs/_partial-conditional-params.mdx'; import PartialVectorSearchResults from '@site/docs/_partial-vector-search-results.mdx'; import PartialVectorStoreInstance from '@site/docs/_partial-vector-store-instance.mdx';

<Icon name="Blocks" aria-hidden="true" /> Bundles contain custom components that support specific third-party integrations with Langflow.

The IBM bundle provides access to IBM watsonx.ai models for text and embedding generation, plus an IBM Db2 Vector Store. These components require an IBM watsonx.ai deployment with API credentials, and/or a reachable IBM Db2 instance with the ibm-db driver.

Install the IBM bundle

The IBM bundle is included in the lfx-ibm Extension bundle, which is installed automatically as part of uv pip install langflow.

If you need to install it separately, run:

bash

uv pip install lfx-ibm
uv run langflow run

To verify the bundle is loaded in your environment:

bash

lfx extension list

IBM watsonx.ai

The IBM watsonx.ai component generates text using supported foundation models in IBM watsonx.ai. To use gateway models, use the OpenAI text generation component with the gateway model's OpenAI-compatible endpoint.

You can use the IBM watsonx.ai component anywhere you need a language model in a flow.

IBM watsonx.ai parameters {#ibm-watsonxai-parameters}

Name	Type	Description
url	String	Input parameter. The watsonx API base URL for your deployment and region.
project_id	String	Input parameter. Your watsonx Project ID.
api_key	SecretString	Input parameter. A watsonx API key to authenticate watsonx API access to the specified watsonx.ai deployment and model.
model_name	String	Input parameter. The name of the watsonx model to use. Options are dynamically fetched from the API.
max_tokens	Integer	Input parameter. The maximum number of tokens to generate. Default: `1000`.
stop_sequence	String	Input parameter. The sequence where generation should stop.
temperature	Float	Input parameter. Controls randomness in the output. Default: `0.1`.
top_p	Float	Input parameter. Controls nucleus sampling, which limits the model to tokens whose probability is below the `top_p` value. Range: Default: `0.9`.
frequency_penalty	Float	Input parameter. Controls frequency penalty. A positive value decreases the probability of repeating tokens, and a negative value increases the probability. Range: Default: `0.5`.
presence_penalty	Float	Input parameter. Controls presence penalty. A positive value increases the likelihood of new topics being introduced. Default: `0.3`.
seed	Integer	Input parameter. A random seed for the model. Default: `8`.
logprobs	Boolean	Input parameter. Whether to return log probabilities of output tokens or not. Default: `true`.
top_logprobs	Integer	Input parameter. The number of most likely tokens to return at each position. Default: `3`.
logit_bias	String	Input parameter. A JSON string of token IDs to bias or suppress.

IBM watsonx.ai output

The IBM watsonx.ai component can output either a Model Response (Message) or a Language Model (LanguageModel).

Use the Language Model output when you want to use an IBM watsonx.ai model as the LLM for another LLM-driven component, such as an Agent or Smart Transform component. For more information, see Language model components.

The LanguageModel output from the IBM watsonx.ai component is an instance of [ChatWatsonx](https://docs.langchain.com/oss/python/integrations/chat/ibm_watsonx) configured according to the component's parameters.

IBM watsonx.ai Embeddings

The IBM watsonx.ai Embeddings component uses the supported foundation models in IBM watsonx.ai for embedding generation.

The output is Embeddings generated with WatsonxEmbeddings.

For more information about using embedding model components in flows, see Embedding model components.

IBM watsonx.ai Embeddings parameters

Name	Display Name	Info
url	watsonx API Endpoint	Input parameter. The watsonx API base URL for your deployment and region.
project_id	watsonx project id	Input parameter. Your watsonx Project ID.
api_key	API Key	Input parameter. A watsonx API key to authenticate watsonx API access to the specified watsonx.ai deployment and model.
model_name	Model Name	Input parameter. The name of the embedding model to use. Supports default embedding models and automatically updates after connecting to your watsonx.ai deployment.
truncate_input_tokens	Truncate Input Tokens	Input parameter. The maximum number of tokens to process. Default: `200`.
input_text	Include the original text in the output	Input parameter. Determines if the original text is included in the output. Default: `true`.

Default embedding models

By default, the IBM watsonx.ai Embeddings component supports the following default models:

sentence-transformers/all-minilm-l12-v2: 384-dimensional embeddings
ibm/slate-125m-english-rtrvr-v2: 768-dimensional embeddings
ibm/slate-30m-english-rtrvr-v2: 768-dimensional embeddings
intfloat/multilingual-e5-large: 1024-dimensional embeddings

After entering your API endpoint and credentials, the component automatically fetches the list of available models from your watsonx.ai deployment.

IBM Db2 Vector Store

You can use the IBM Db2 Vector Store component to read and write to an IBM Db2 database using an instance of DB2VS vector store. Includes support for remote Db2 instances with enterprise-grade security and performance.

<details> <summary>About vector store instances</summary> <PartialVectorStoreInstance /> </details>

When writing, the component can create a new table at the specified location.

:::tip IBM Db2 Vector Store provides enterprise-grade vector search capabilities with built-in security validation and support for multiple distance strategies. :::

Use the IBM Db2 Vector Store component in a flow

The IBM Db2 Vector Store component can be used for both reads and writes:

When writing, it splits JSON from a URL component into chunks, computes embeddings with attached Embedding Model component, and then loads the chunks and embeddings into the Db2 vector store. To trigger writes, click <Icon name="Play" aria-hidden="true"/> Run component on the IBM Db2 Vector Store component.
When reading, it uses chat input to perform a similarity search on the vector store, and then print the search results to the chat. To trigger reads, open the Playground and enter a chat message.

After running the flow once, you can click <Icon name="TextSearch" aria-hidden="true"/> Inspect Output on each component to understand how the data transformed as it passed from component to component.

IBM Db2 Vector Store parameters

You can inspect a vector store component's parameters to learn more about the inputs it accepts, the features it supports, and how to configure it.

For information about accepted values and functionality, see the provider's documentation or inspect component code.

Name	Type	Description
Table Name (`collection_name`)	String	Input parameter. The name of your Db2 table to store vectors. Default: `LANGFLOW_VECTORS`. The table will be created if it doesn't exist.
Database Name (`database`)	String	Input parameter. Name of the Db2 database. Use a Generic-typed global variable or direct input. Credential-typed variables are not allowed for database names.
Hostname (`hostname`)	String	Input parameter. Db2 server hostname or IP address. Use a Generic-typed global variable or direct input.
Port (`port`)	Integer	Input parameter. Db2 server port. Default: `50000`.
Username (`username`)	String	Input parameter. Db2 database username. Use a Generic-typed global variable or direct input.
Password (`password`)	String	Input parameter. Db2 database password. This should use a Credential-typed global variable for security.
Ingest Data (`ingest_data`)	JSON or Table	Input parameter. `JSON` or `Table` input containing the records to write to the vector store. Only relevant for writes.
Search Query (`search_query`)	String	Input parameter. The query to use for vector search. Only relevant for reads.
Cache Vector Store (`should_cache_vector_store`)	Boolean	Input parameter. If `true`, the component caches the vector store in memory for faster reads. Default: Enabled (`true`).
Embedding (`embedding`)	Embeddings	Input parameter. The embedding function to use for the vector store. You must attach an Embedding Model component to generate embeddings for your data.
Allow Duplicates (`allow_duplicates`)	Boolean	Input parameter. If `true` (default), writes don't check for existing duplicates in the collection, allowing you to store multiple copies of the same content. If `false`, writes won't add documents that match existing documents already present in the collection. Only relevant for writes.
Search Type (`search_type`)	String	Input parameter. The type of search to perform: `Similarity`, `MMR`, or `similarity_score_threshold`. Only relevant for reads.
Number of Results (`number_of_results`)	Integer	Input parameter. The number of search results to return. Default: `4`. Only relevant for reads.
Distance Strategy (`distance_strategy`)	String	Input parameter. Distance calculation strategy: `COSINE`, `EUCLIDEAN_DISTANCE`, or `DOT_PRODUCT`. Default: `COSINE`.