fastsafetensors
High-performance safetensors model loader
Decision gist · record as of 2026-08-14
Yes, if you load large safetensors models and want faster initialization. The package is actively maintained, has no known vulnerabilities, and offers substantial speedups with minimal friction—install and use the API or CLI. Medium install friction is standard for compiled packages with architecture-specific wheels.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10–3.14; compiled wheels are architecture-specific (x86_64, aarch64, Windows); performance gains depend on platform and storage backend.
- Medium install friction due to compiled wheels for specific Python versions (3.10–3.14) and architectures (x86_64, aarch64, Windows).
- Active maintenance with recent commits and no known vulnerabilities.
License · maintenance · safety
Apache-2.0 (permissive) — Apache-2.0 permissive license allows commercial and private use with minimal restrictions; you must retain license notices in distributions.
last release 2026-07-07 (38 days) · last repo commit 2026-08-14 · 163 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 2,095,638 downloads/mo, #3,299 on PyPI
Alternatives
Verify before relying
pip install fastsafetensors
import fastsafetensors
# Use fastsafetensors APIs directly (see docs/overview.md)- Whether PyTorch 2.11.0 is required or if other versions are compatible despite CI testing only that version.
- Whether GDS (GPU Direct Storage) is required for documented performance gains or if fallback paths work without it.
- Whether typer is used for CLI tooling or is a transitive dependency.
What it is and what it does
fastsafetensors is a high-performance loader for safetensors model files, designed to accelerate model initialization in machine learning frameworks. It replaces the default safetensors deserializer with optimized I/O routines that exploit GPU and storage hardware capabilities—including NVIDIA GDS, AMD ROCm, and NVMe—to reduce model loading time. The library integrates into existing tools as a command-line option.
The package provides both a direct Python API and CLI integration via typer. It supports Linux/CUDA, ROCm, Windows, and specialized storage systems like 3FS and unified-memory architectures. Installation requires a precompiled wheel for your Python version and architecture.
Use it for
- Reduce startup time when serving large language models by using fastsafetensors as the model loader.
- Speed up model initialization in custom Python code that loads large safetensors files from local or remote storage.
- Optimize model loading on AMD ROCm systems or systems with GPU Direct Storage to maximize NVMe throughput.
- Integrate into machine learning inference pipelines where model load time is a bottleneck.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you load large safetensors models and want faster initialization.
The package is actively maintained, has no known vulnerabilities, and offers substantial speedups with minimal friction—install and use the API or CLI. Medium install friction is standard for compiled packages with architecture-specific wheels.
Install
fastsafetensors on PyPI
Before you install
Medium install friction due to compiled wheels for specific Python versions (3.10–3.14) and architectures (x86_64, aarch64, Windows). Active maintenance with recent commits and no known vulnerabilities.
Requires Python 3.10–3.14; compiled wheels are architecture-specific (x86_64, aarch64, Windows); performance gains depend on platform and storage backend.
License in practice
Apache-2.0 permissive license allows commercial and private use with minimal restrictions; you must retain license notices in distributions.
Quickstart
pip install fastsafetensors
import fastsafetensors
# Use fastsafetensors APIs directly (see docs/overview.md)Verify before relying
- Whether PyTorch 2.11.0 is required or if other versions are compatible despite CI testing only that version.
- Whether GDS (GPU Direct Storage) is required for documented performance gains or if fallback paths work without it.
- Whether typer is used for CLI tooling or is a transitive dependency.
Package facts
| License | Apache-2.0 permissive |
| Python support | Supports the current Python release <3.15,>=3.10 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 1 packagetyper |
| Maintenance | Actively maintained 38 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 2,095,638 / month, #3,299 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Programming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.13Programming Language :: Python :: 3.14 |
Evidence: fastsafetensors-0.3.3-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; fastsafetensors-0.3.3-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; fastsafetensors-0.3.3-cp310-cp310-win_amd64.whl; fastsafetensors-0.3.3-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; fastsafetensors-0.3.3-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; fastsafetensors-0.3.3-cp311-cp311-win_amd64.whl; fastsafetensors-0.3.3-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; fastsafetensors-0.3.3-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; fastsafetensors-0.3.3-cp312-cp312-win_amd64.whl; fastsafetensors-0.3.3-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; fastsafetensors-0.3.3-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; fastsafetensors-0.3.3-cp313-cp313-win_amd64.whl; fastsafetensors-0.3.3-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; fastsafetensors-0.3.3-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; fastsafetensors-0.3.3-cp314-cp314-win_amd64.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 › “fast safetensors loader”
- fastsafetensorsLoads safetensors model files significantly faster than the standard…
- instanttensorInstantTensor provides a high-throughput Safetensors loader optimized…
- spandrelSpandrel loads and runs pre-trained PyTorch models by automatically…
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 instanttensor · safetensors · compressed-tensors · mmgp · sgl-kernel · cutensornet-cu13 · sglang · nvidia-cusparse-cu12 · cutensor-cu12 · nvidia-cutlass-dsl-libs-base