What is PromptTemplate?
PromptTemplate is the foundational string-formatting primitive in LangChain. It wraps a Python f-string-style template with curly-brace placeholders and validates that all required input variables are supplied before formatting. Unlike Python's str.format(), PromptTemplate raises a validation error if a required variable is missing—immediately rather than silently producing a broken prompt.
PromptTemplate works with any model that accepts a single string input. For chat models (ChatOpenAI, ChatAnthropic), prefer ChatPromptTemplate, which produces structured message lists. PromptTemplate.from_template() is the quickest constructor—it infers input_variables from the curly-brace placeholders automatically, so you don't need to list them manually.
Partial variables let you freeze some inputs and leave others open. partial() returns a new PromptTemplate with the supplied variables pre-filled—useful when the system context is fixed but the user query changes per invocation. You can also pass a callable as a partial variable, and LangChain will call it at format time.
When to Use
You need a simple template with variables. Use PromptTemplate for string-based templates.
Use Cases
- • Simple text prompts
- • Template reusability
- • Variable insertion
- • Prompt composition
- • Few-shot templates
- • Dynamic prompts
Key Features
- ✓ Simple {variable} syntax
- ✓ Validation
- ✓ Composable
- ✓ Format support
- ✓ Reusable
- ✓ LCEL-compatible
When NOT to Use
For chat models with roles—use ChatPromptTemplate.
Notes
Use ChatPromptTemplate for chat models
PromptTemplate produces a plain string, which is wrapped in a HumanMessage automatically when passed to a chat model. For precise system/user/assistant role control, use ChatPromptTemplate.from_messages() instead.
from_template infers input_variables
PromptTemplate.from_template("Hello {name}") automatically sets input_variables=["name"]. Only pass input_variables explicitly when some curly braces should not be treated as placeholders—escape them as {{ }}.
Partial variables for shared context
template.partial(context="...") binds one variable while leaving others open. The result is a new PromptTemplate—the original is unchanged. You can chain .partial() calls and pass a callable to compute the value at format time.
Template format: f-string vs jinja2
Default template_format="f-string". Switch to template_format="jinja2" to use Jinja2 syntax ({% if %}, {% for %}) for conditional prompts. Escape literal braces as {{ }} in Jinja2 templates to avoid conflicts with variable syntax.
Import
from langchain_core.prompts import PromptTemplateKey Parameters
| Parameter | Type | Default | Purpose |
|---|---|---|---|
| template | str | None | Template string with {variables} |
| input_variables | List[str] | None | List of variable names |
Code Examples
Simple Template
template = 'Tell me a {topic} joke'
pt = PromptTemplate(template=template, input_variables=['topic'])
result = pt.invoke({'topic': 'programming'})Partial Variables
base = PromptTemplate.from_template('You are a {role}. Answer: {question}')
# Freeze the role, leave question open
expert_pt = base.partial(role='Python expert')
result = expert_pt.invoke({'question': 'What is a generator?'})Jinja2 Conditional Template
from langchain_core.prompts import PromptTemplate
# Jinja2 allows conditionals
pt = PromptTemplate(
template='Answer in {lang}.{% if formal %} Be formal.{% endif %} {{question}}',
input_variables=['lang', 'formal', 'question'],
template_format='jinja2'
)Common Mistakes
❌ Not matching variables in template and input_variables
✅ Ensure every {var} in template is in input_variables
Alternatives
| Class | When to Use |
|---|---|
| ChatPromptTemplate | For chat models with role-based messages |
Related LangChain References
Browse the full LangChain API reference index to explore more classes, methods, and decorators, or start with the LangChain introduction tutorial for end-to-end context on building with PromptTemplate and the wider framework.
PromptTemplate FAQ
What is PromptTemplate in LangChain?
Create reusable text prompts with variable placeholders. PromptTemplate is the foundational string-formatting primitive in LangChain. It wraps a Python f-string-style template with curly-brace placeholders and validates that all required input variables are supplied before formatting. Unlike Python's str.format(), PromptTemplate raises a validation error if a required variable is missing—immediately rather than silently producing a broken prompt. PromptTemplate works with any model that accepts a single string input. For chat mode…
Which package provides PromptTemplate?
DevShelfHub documents PromptTemplate from the langchain-core package. Pin your installed LangChain version and match imports to the snippet on this page.
When should I use PromptTemplate?
You need a simple template with variables. Use PromptTemplate for string-based templates.
When should I avoid using PromptTemplate?
For chat models with roles—use ChatPromptTemplate.
How do I import PromptTemplate in Python?
from langchain_core.prompts import PromptTemplate
Where can I explore more LangChain API reference pages?
Open the LangChain API reference index on DevShelfHub to browse classes, methods, and decorators, each with runnable examples, parameters, common mistakes, and cross-links.