onnxruntime-genai
ONNX Runtime GenAI
Decision gist · record as of 2026-08-14
Yes, if you have ONNX-converted models and want a straightforward Python interface to run them on-device or in the cloud. The library is actively maintained, has no known vulnerabilities, and offers broad platform coverage. Install only if you already have or plan to convert models to ONNX format; it is not a model hub or download tool.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires numpy and onnxruntime as runtime dependencies; model files and platform-specific ONNX Runtime support are prerequisites.
- Medium install friction due to pre-built wheels for multiple Python versions and platforms.
- Active maintenance with a release 8 days old.
License · maintenance · safety
MIT License (permissive) — MIT License permits commercial and private use with minimal restrictions; suitable for most projects.
last release 2026-08-06 (8 days)
0 known vulnerabilities (OSV.dev, 2026-08-14) · 131,825 downloads/mo, #11,573 on PyPI
Alternatives
Verify before relying
pip install onnxruntime-genai
import onnxruntime_genai
# Load and run a model via the ONNX Runtime GenAI API- Specific model formats and sizes supported by this version
- Performance characteristics (latency, throughput) on different hardware
- Whether quantization or optimization tooling is included
- API stability and backward compatibility guarantees
- Python version requirements (requires_python field is unspecified)
What it is and what it does
ONNX Runtime GenAI is a Python library that wraps ONNX Runtime to simplify inference of language models and multi-modal models. It abstracts away the complexity of loading ONNX model files and managing inference sessions, letting you focus on running models rather than configuring the runtime. The library depends on numpy and onnxruntime, and is distributed as pre-built wheels across Windows, Linux, and macOS on both x86 and ARM architectures.
The package is intended for scenarios where you want to deploy or experiment with LLMs and SLMs locally or in cloud environments without relying on external APIs. It bridges the gap between raw ONNX Runtime (a low-level inference engine) and high-level model frameworks, offering a middle ground for developers who have ONNX-converted models and want a straightforward Python interface to run them.
Use it for
- Deploy a quantized LLM locally on a laptop or edge device for inference without cloud API calls
- Run multi-modal models in a Python application using pre-converted ONNX weights
- Integrate on-device SLM inference into a web service or backend for low-latency responses
- Experiment with different model architectures by swapping ONNX model files without code changes
- Build a chatbot or text-generation pipeline that runs entirely on-premises for privacy
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you have ONNX-converted models and want a straightforward Python interface to run them on-device or in the cloud.
The library is actively maintained, has no known vulnerabilities, and offers broad platform coverage. Install only if you already have or plan to convert models to ONNX format; it is not a model hub or download tool.
Install
onnxruntime-genai on PyPI
Before you install
Medium install friction due to pre-built wheels for multiple Python versions and platforms. Active maintenance with a release 8 days old.
Requires numpy and onnxruntime as runtime dependencies; model files and platform-specific ONNX Runtime support are prerequisites.
License in practice
MIT License permits commercial and private use with minimal restrictions; suitable for most projects.
Quickstart
pip install onnxruntime-genai
import onnxruntime_genai
# Load and run a model via the ONNX Runtime GenAI APIVerify before relying
- Specific model formats and sizes supported by this version
- Performance characteristics (latency, throughput) on different hardware
- Whether quantization or optimization tooling is included
- API stability and backward compatibility guarantees
- Python version requirements (requires_python field is unspecified)
Package facts
| License | MIT License permissive |
| Python support | Not specified |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 2 packagesnumpyonnxruntime |
| Maintenance | Actively maintained 8 days since the last release |
| First released | |
| Downloads | 131,825 / month, #11,573 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
Evidence: onnxruntime_genai-0.15.2-cp311-cp311-macosx_12_0_arm64.whl; onnxruntime_genai-0.15.2-cp311-cp311-manylinux_2_28_aarch64.whl; onnxruntime_genai-0.15.2-cp311-cp311-manylinux_2_28_x86_64.whl; onnxruntime_genai-0.15.2-cp311-cp311-win_amd64.whl; onnxruntime_genai-0.15.2-cp311-cp311-win_arm64.whl; onnxruntime_genai-0.15.2-cp312-cp312-macosx_12_0_arm64.whl; onnxruntime_genai-0.15.2-cp312-cp312-manylinux_2_28_aarch64.whl; onnxruntime_genai-0.15.2-cp312-cp312-manylinux_2_28_x86_64.whl; onnxruntime_genai-0.15.2-cp312-cp312-win_amd64.whl; onnxruntime_genai-0.15.2-cp312-cp312-win_arm64.whl; onnxruntime_genai-0.15.2-cp313-cp313-macosx_12_0_arm64.whl; onnxruntime_genai-0.15.2-cp313-cp313-manylinux_2_28_aarch64.whl; onnxruntime_genai-0.15.2-cp313-cp313-manylinux_2_28_x86_64.whl; onnxruntime_genai-0.15.2-cp313-cp313-win_amd64.whl; onnxruntime_genai-0.15.2-cp313-cp313-win_arm64.whl; onnxruntime_genai-0.15.2-cp314-cp314-macosx_12_0_arm64.whl; onnxruntime_genai-0.15.2-cp314-cp314-manylinux_2_28_aarch64.whl; onnxruntime_genai-0.15.2-cp314-cp314-manylinux_2_28_x86_64.whl; onnxruntime_genai-0.15.2-cp314-cp314-win_amd64.whl; onnxruntime_genai-0.15.2-cp314-cp314-win_arm64.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “run LLM on device”
- onnxruntime-genaiRuns small and large language models and multi-modal models on-device…
- vllm-cpuvllm-cpu provides CPU-optimized inference for large language models…
- litserveLitServe lets you build custom inference servers in Python by…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also onnxruntime-openvino · litert-lm-builder · optimum-onnx · tflite-runtime · ai-edge-litert · optimum · onnxslim · executorch · openvino-genai · genai-perf