Opper Python SDK
Python client for the Opper API.
Install
pip install opperai
Quick Start
from opperai import Opper
opper = Opper() # uses OPPER_API_KEY env var
result = opper.call("summarize", input={"text": "Long article..."})
print(result.data)
# Stream a function
for chunk in opper.stream("summarize", input={"text": "Long article..."}):
if chunk.type == "content":
print(chunk.delta, end="")
if chunk.type == "complete":
print(chunk.data)
Schema Support
Pass Pydantic models, dataclasses, TypedDicts, or raw JSON Schema dicts for input_schema and output_schema — the SDK resolves them to JSON Schema automatically.
from pydantic import BaseModel
class Summary(BaseModel):
summary: str
entities: list[str]
result = opper.call(
"extract",
input={"text": "Marie Curie was a physicist in Paris."},
output_schema=Summary,
)
result.data.summary # str — typed!
result.data.entities # list[str]
Dataclasses, TypedDicts, and plain dicts also work. See 01a_using_schemas.py and 01b_using_other_schemas.py.
Observability
Use trace() as a decorator or context manager to group calls under a single trace span. Nesting works naturally.
@opper.trace("my-pipeline")
def run():
a = opper.call("step-1", input="hello")
b = opper.call("step-2", input=a.data)
# or as a context manager
with opper.trace("my-pipeline") as span:
opper.call("step-1", input="hello")
Agent SDK
Build AI agents with tool use, streaming, multi-agent composition, and MCP integration.
from opperai import Agent, tool
@tool
def get_weather(city: str) -> str:
"""Get the current weather for a city."""
return f"Sunny, 22°C in {city}"
agent = Agent(
name="weather-assistant",
instructions="You are a helpful weather assistant.",
tools=[get_weather],
)
# Run — get the final result
result = await agent.run("What's the weather in Paris?")
print(result.output)
print(result.meta.usage) # token usage across all iterations
# Stream — observe events as the agent works
stream = agent.stream("What's the weather in Paris?")
async for event in stream:
if event.type == "text_delta":
print(event.text, end="", flush=True)
if event.type == "tool_start":
print(f"\nCalling {event.name}...")
result = await stream.result()
Structured Output
from pydantic import BaseModel
class Sentiment(BaseModel):
label: str
score: float
agent = Agent(
name="analyzer",
instructions="Analyze the sentiment of the input.",
output_schema=Sentiment,
)
result = await agent.run("I love this product!")
result.output.label # str — typed via Pydantic
result.output.score # float
Multi-Agent Composition
researcher = Agent(name="researcher", instructions="...", tools=[web_search])
writer = Agent(
name="writer",
instructions="Write clear reports using research.",
tools=[researcher.as_tool(name="research", description="Research a topic")],
)
result = await writer.run("Write a report on AI agents")
MCP Integration
from opperai.agent.mcp import mcp, MCPStdioConfig
agent = Agent(
name="file-assistant",
instructions="Help users manage files.",
tools=[mcp(MCPStdioConfig(name="fs", command="uvx", args=["mcp-server-filesystem", "/tmp"]))],
)
Conversation (Multi-Turn)
conversation = agent.conversation()
r1 = await conversation.send("My name is Alice")
r2 = await conversation.send("What is my name?")
# r2.output → "Your name is Alice"
Examples
| # | Example | What it shows |
|---|---|---|
| 00 | First call | Simplest possible call |
| 01a | Pydantic schemas | Type-safe output with Pydantic |
| 01b | Other schemas | Dataclass, TypedDict, raw dict |
| 02 | Streaming | Stream deltas + complete event |
| 03a | Tools (call) | Tool definitions with call() |
| 03b | Tools (stream) | Tool call chunks in streaming |
| 04a | Generate image | Image generation |
| 04b | Describe image | Vision / image description |
| 04c | Edit image | Image editing |
| 05 | Audio | Text-to-speech + speech-to-text |
| 06 | Video | Video generation |
| 07 | Embeddings | Vector embeddings + similarity |
| 08 | Function mgmt | List, get, revisions, delete |
| 09 | Observability | Tracing with decorator + context manager |
| 09b | Manual tracing | Manual span creation |
| 09c | Traces | List, get, and inspect traces |
| 10 | Models | List available models |
| 11 | Realtime | Mint a ticket for browser-direct voice WebSocket |
| 12 | Knowledge base | Semantic search with knowledge bases |
| 13 | Web tools | Web search and URL fetch (beta) |
Run a single example:
export OPPER_API_KEY="your-key"
uv run python examples/getting-started/00_your_first_call.py
Run all examples:
uv run python examples/run_all.py
Configuration
| Parameter | Default | Env Var |
|---|---|---|
api_key |
— | OPPER_API_KEY |
base_url |
https://api.opper.ai |
OPPER_BASE_URL |
headers |
{} |
— |
Error Handling
from opperai import ApiError
try:
opper.call("my-fn", input="hello")
except ApiError as e:
print(e.status, e.body)
Async Support
All methods have _async variants:
result = await opper.call_async("summarize", input={"text": "..."})
async for chunk in opper.stream_async("summarize", input={"text": "..."}):
print(chunk.delta, end="")
Requirements
- Python 3.10+
- Optional:
pip install opperai[pydantic]for Pydantic schema support
License
MIT
1 maintainer
Metadata
1 maintainer
Release files for opperai 2.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| opperai-2.1.1.tar.gz | 2.1 MB | With provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| opperai-2.1.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 2.2 MB
Release files / opperai-2.1.1.tar.gz
| Download URL | opperai-2.1.1.tar.gz |
|---|---|
| Size | 2.1 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
f728733aaeb1920f4ad9eb136cae44ab3f1ae1ea940f7fc4e67230250f8faf05
|
|
BLAKE2b-256 checksum How to use checksums |
7999befc7eec98e75b577b7e9fbc0495ad721ef70b4547c73db00e981a3841c2
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 25, 2026.
Transparency logRelease files / opperai-2.1.1-py3-none-any.whl
| Download URL | opperai-2.1.1-py3-none-any.whl |
|---|---|
| Size | 70.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
eb9b9ae55f3464f9191e122f15816f771789d114c4c2e70e42bfe67dd0e106f4
|
|
BLAKE2b-256 checksum How to use checksums |
9963207a77c04c3bbe60a507a21be5e7e1905f6be337675694b134f518e8b558
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 25, 2026.
Transparency logSecurity for opperai 2.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 2 files in this release have been uploaded with provenance metadata.
Attestation for opperai-2.1.1.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jun 25, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
