# Use OpenAI Models
Use OpenAI models in your pipelines.
***
## About This Task
You can use OpenAI's embedding models and LLMs:
- For a list of embedding models, see [OpenAI documentation](https://platform.openai.com/docs/guides/embeddings/embedding-models).
- For a list of LLMs, see [OpenAI model overview](https://platform.openai.com/docs/models/models-overview).
## Prerequisites
You need an API key from an active OpenAI account. For details on obtaining it, see [Secret keys in OpenAI](https://help.openai.com/en/articles/4936850-where-do-i-find-my-openai-api-key).
## Use OpenAI Models
First, connect to OpenAI through the Integrations page. You can set up the connection for a single workspace or for the whole organization:
Then, add a component that uses an OpenAI model to your pipeline. Here are the components by the model type they use:
- Embedding models:
- [`OpenAITextEmbedder`](/docs/reference/pipeline-components/integrations/openai/OpenAITextEmbedder.mdx): Calculates embeddings for text, like query. Often used in query pipelines to embed a query and pass the embedding to an embedding retriever.
- [`OpenAIDocumentEmbedder`](/docs/reference/pipeline-components/integrations/openai/OpenAIDocumentEmbedder.mdx): Calculates embeddings for documents. Often used in indexes to embed documents and pass them to `DocumentWriter`.
- LLMs:
- [`OpenAIChatGenerator`](/docs/reference/pipeline-components/legacy-components/OpenAIChatGenerator.mdx): Generates text using OpenAI models, often used in RAG pipelines.
- [`LLM`](/docs/reference/pipeline-components/ai/LLM.mdx): Generates text using OpenAI models, often used in RAG pipelines.
## Available OpenAI LLMs
The following OpenAI LLMs are available in :
| Model | Description |
|-------|-------------|
| GPT-5.6 Sol | Latest GPT-5.6 variant with `reasoning_effort` and `reasoning_summary` parameters. |
| GPT-5.6 Terra | Latest GPT-5.6 variant with `reasoning_effort` and `reasoning_summary` parameters. |
| GPT-5.6 Luna | Latest GPT-5.6 variant with `reasoning_effort` and `reasoning_summary` parameters. |
| GPT-5.5 | GPT-5.5 with `reasoning_effort` and `reasoning_summary` parameters. |
| GPT-5.5 Pro | GPT-5.5 Pro variant. |
| GPT-5.4 | GPT-5.4 model. |
| GPT-5.4 Pro | GPT-5.4 Pro variant. |
| GPT-5.4 mini | Smaller, faster GPT-5.4 variant. |
| GPT-5.4 nano | Compact GPT-5.4 variant. |
| GPT-5.2 | GPT-5.2 model. |
| GPT-5.2 Pro | GPT-5.2 Pro variant (no text-to-speech support). |
| GPT-5.1 | GPT-5.1 model. |
| GPT-5 | GPT-5 base model. |
| GPT-5 Pro | GPT-5 Pro variant. |
| GPT-5 mini | Smaller, faster GPT-5 variant. |
| GPT-5 nano | Compact GPT-5 variant. |
| GPT-4.1 | GPT-4.1 model. |
| GPT-4.1 mini | Smaller GPT-4.1 variant. |
| GPT-4.1 nano | Compact GPT-4.1 variant. |
| GPT-4o | GPT-4o model. |
| GPT-4o mini | Smaller GPT-4o variant. |
| GPT-3.5 Turbo | GPT-3.5 Turbo model. |
## Usage Examples
This is an example of how to use OpenAI's embedding models and an LLM in an index and a query pipeline (each in a separate tab):
```yaml
components:
# ...
splitter:
type: haystack.components.preprocessors.document_splitter.DocumentSplitter
init_parameters:
split_by: word
split_length: 250
split_overlap: 30
document_embedder:
type: haystack.components.embedders.openai_document_embedder.OpenAIDocumentEmbedder
init_parameters:
model: text-embedding-ada-002 # the model to use
writer:
type: haystack.components.writers.document_writer.DocumentWriter
init_parameters:
document_store:
type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
init_parameters:
embedding_dim: 768
similarity: cosine
policy: OVERWRITE
connections: # Defines how the components are connected
# ...
- sender: splitter.documents
receiver: document_embedder.documents
- sender: document_embedder.documents
receiver: writer.documents
```
```yaml Query Pipeline
# haystack-pipeline
pipeline_output_type: "chat"
components:
retriever:
# Selects the most similar documents from the document store
type: haystack_integrations.components.retrievers.opensearch.open_search_hybrid_retriever.OpenSearchHybridRetriever
init_parameters:
document_store:
type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
init_parameters:
embedding_dim: 768
index: Standard-Index-English
max_chunk_bytes: 104857600
return_embedding: false
create_index: true
settings:
index.knn: true
top_k: 20 # The number of results to return
embedder:
type: haystack.components.embedders.openai_text_embedder.OpenAITextEmbedder
init_parameters:
model: text-embedding-ada-002
fuzziness: 0
ranker:
type: haystack_integrations.components.rankers.sentence_transformers.SentenceTransformersSimilarityRanker
init_parameters:
model: intfloat/simlm-msmarco-reranker
top_k: 8
device: null
meta_field_grouping_ranker:
type: haystack.components.rankers.meta_field_grouping_ranker.MetaFieldGroupingRanker
init_parameters:
group_by: file_id
subgroup_by:
sort_docs_by: split_id
qa_llm:
type: haystack.components.generators.chat.llm.LLM
init_parameters:
# You can swap this for any other model. Switch to the Builder view and choose another model from the list on the component card.
chat_generator:
init_parameters:
model: gpt-5.4
type: haystack.components.generators.chat.openai_responses.OpenAIResponsesChatGenerator
system_prompt:
user_prompt: >-
{% message role="user" %}
You are a technical expert.
You answer questions truthfully based on provided documents.
Ignore typing errors in the question.
For each document check whether it is related to the question.
Only use documents that are related to the question to answer it.
Ignore documents that are not related to the question.
If the answer exists in several documents, summarize them.
Only answer based on the documents provided. Don't make things up.
Just output the structured, informative and precise answer and nothing
else.
If the documents can't answer the question, say so.
Always use references in the form [NUMBER OF DOCUMENT] when using
information from a document, e.g. [3] for Document [3] .
Never name the documents, only enter a number in square brackets as a
reference.
The reference must only refer to the number that comes in square
brackets after the document.
Otherwise, do not use brackets in your answer and reference ONLY the
number of the document without mentioning the word document.
These are the documents:
{%- if documents|length > 0 %}
{% for document in documents %}
Document [{{ loop.index }}] :
Name of Source File: {{ document.meta.file_name }}
{{ document.content }}
{% endfor %}
{%- else %}
No relevant documents found.
Respond with "Sorry, no matching documents were found, please adjust the
filters or try a different question."
{% endif %}
Question: {{ question.text }}
Answer:
{% endmessage %}
required_variables: "*"
streaming_callback:
connections:
- sender: retriever.documents
receiver: ranker.documents
- sender: ranker.documents
receiver: meta_field_grouping_ranker.documents
- sender: meta_field_grouping_ranker.documents
receiver: qa_llm.documents
inputs:
filters:
- retriever.filters_bm25
- retriever.filters_embedding
files: []
query:
- qa_llm.question
- ranker.query
- retriever.query
outputs:
documents: meta_field_grouping_ranker.documents
messages: qa_llm.messages
updated_query:
max_runs_per_component: 100
metadata: {}
```
Here is how to connect the components in Pipeline Builder. In the index, `OpenAIDocumentEmbedder` receives documents from `DocumentSplitter` and then passes the embedded documents to `DocumentWriter`, which writes them into the Document Store:
In a query pipeline, `OpenAITextEmbedder` embeds the query using the same model as the `OpenAIDocumentEmbedder` in the index. Then, it sends the embedded query to the retriever, which fetches matching documents and sends them to `PromptBuilder`. The `LLM` component uses the rendered prompt to generate a reply and sends it to `AnswerBuilder`, which builds a `GeneratedAnswer` object.
## Related Information
- [Add Custom Model Definitions](/docs/how-to-guides/managing-access/add-custom-model-definitions.mdx)
- [LLM](/docs/reference/pipeline-components/ai/LLM.mdx)