// event · sdk
openai-python 3.32.0 adds cache settings, prompt-based voice creation, and Live session Decisions
sdkvendor: openaipublished: 2h agoconfidence: 0.90id: 8b27af0c-0f2e-4103-b2d4-b919e7da0731
The OpenAI Python SDK v3.32.0 adds cache settings and key options, prompt-based voice creation, regional processing residency options, response error diagnostics and model notices, audio fixture playback in Live sessions, and Decisions support in Live sessions. Also fixes Realtime response error message declarations and refactors shared WebSocket connection handling.
capability delta
An agent can now create custom voices from text prompts, pin requests to regional processing residency, configure cache keys/settings, and use Decisions plus audio fixture playback inside Live/Realtime sessions.
curator note
Multiple user-facing API features in one SDK bump: voice creation from prompts and Live-session Decisions are the most novel agent capabilities.
// install
pip install openai==3.32.0pypi · openai @ 3.32.0
// example
python
from openai import OpenAI
client = OpenAI()
# Prompt-based voice creation (new in 3.32.0)
voice = client.audio.voices.create(
name="calm-narrator",
prompt="A soft-spoken middle-aged narrator with a warm, measured tone.",
)
# Use regional processing residency + cache settings on a response
resp = client.responses.create(
model="gpt-4.1",
input="Summarize today's standup notes.",
extra_body={
"processing": {"residency": "eu"},
"cache": {"key": "standup-2026-10-16", "ttl_seconds": 3600},
},
)
print(resp.output_text)// links