PyPI page
Home page
Author:
Alex Margarit
License:
Apache-2.0
Summary:
The production backend for self-hosted agents — the OpenAI Responses API with server-side conversation state (durable with Redis), universal tool calling, and reasoning, alongside embeddings, speech, and image generation, behind one OpenAI-compatible endpoint. Built on Ray Serve.
Latest version:
0.8.1
Required dependencies:
argparse
|
asyncio
|
cryptography
|
fastapi
|
httpx
|
httpx2
|
huggingface-hub
|
mcp
|
numpy
|
protobuf
|
psutil
|
pydantic
|
pydantic-yaml
|
python-multipart
|
ray
|
redis
|
requests
Optional dependencies:
bitsandbytes
|
diffusers
|
fakeredis
|
flashinfer-python
|
librosa
|
mkdocs-material
|
numba
|
nvidia-ml-py
|
onnxruntime
|
onnxruntime-gpu
|
opentelemetry-exporter-otlp
|
opentelemetry-sdk
|
pre-commit
|
pyright
|
pytest
|
pytest-asyncio
|
pywhispercpp
|
ruff
|
scipy
|
sherpa-onnx
|
sherpa-onnx-core
|
soundfile
|
stable-diffusion-cpp-python
|
torch
|
torchaudio
|
torchvision
|
transformers
|
vllm
Downloads last day:
7
Downloads last week:
225
Downloads last month:
133
7-day smoothing