fastsafetensors

fastsafetensors is an efficient safetensors loader. If you develop your own code that loads large safetensors files, you can try fastsafetensors APIs (see docs). For example, vLLM and SGLang have --load-format fastsafetensors command-line argument to speed up their initialization.

This library supports Linux/CUDA, ROCm, Windows, 3FS, unified-memory systems such as DGX Spark, and so on. We welcome more platform/storage-specific optimizations like them by adding new copier backends. Our CI tests Python 3.10-3.14 with PyTorch 2.11.0.

Performance Highlights

Performance highlights from the CLOUD 2025 paper and benchmark docs:

Standalone model loading was 4.8x-7.5x faster than the default safetensors deserializer on Llama, Falcon, and Bloom models, and reached 26.4 GB/s NVMe read throughput for Llama-70B on four GPUs with GDS.
In the paper's vLLM integration experiment, startup time dropped from 12.39s to 4.74s for Llama-2-13B on 4x L40S GPUs, and from 16.04s to 6.88s on 1x A100.
On AMD ROCm without GDS, the documented nogds path reached 6.02 GB/s for GPT-2 Medium versus 1.28 GB/s with mmap (4.7x throughput), and 2.62 GB/s for GPT-2 versus 1.01 GB/s with mmap (2.6x throughput). See the report for more details.

Quick Start

pip install fastsafetensors
pip install vllm # for quick demo
vllm serve Qwen/Qwen3-0.6B --load-format fastsafetensors
...
Loading safetensors using Fastsafetensor loader:   0% Completed | 0/1 [00:00<?, ?it/s]
Loading safetensors using Fastsafetensor loader: 100% Completed | 1/1 [00:00<00:00,  1.23it/s]

Design Details

See Overview for features, basic API usage, and configuration.

Code of Conduct

Please refer to Foundation Model Stack Community Code of Conduct.

Development

See Development.

Publication

Takeshi Yoshimura, Tatsuhiro Chiba, Manish Sethi, Daniel Waddington, Swaminathan Sundararaman. (2025) Speeding up Model Loading with fastsafetensors arXiv:2505.23072 and IEEE CLOUD 2025.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fastsafetensors-0.3.3.tar.gz (77.8 kB view details)

Uploaded Jul 7, 2026 Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

fastsafetensors-0.3.3-cp314-cp314-win_amd64.whl (435.7 kB view details)

Uploaded Jul 7, 2026 CPython 3.14Windows x86-64

fastsafetensors-0.3.3-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.14manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

fastsafetensors-0.3.3-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.14manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

fastsafetensors-0.3.3-cp313-cp313-win_amd64.whl (425.0 kB view details)

Uploaded Jul 7, 2026 CPython 3.13Windows x86-64

fastsafetensors-0.3.3-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.13manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

fastsafetensors-0.3.3-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.13manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

fastsafetensors-0.3.3-cp312-cp312-win_amd64.whl (424.9 kB view details)

Uploaded Jul 7, 2026 CPython 3.12Windows x86-64

fastsafetensors-0.3.3-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.12manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

fastsafetensors-0.3.3-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.12manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

fastsafetensors-0.3.3-cp311-cp311-win_amd64.whl (424.3 kB view details)

Uploaded Jul 7, 2026 CPython 3.11Windows x86-64

fastsafetensors-0.3.3-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.11manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

fastsafetensors-0.3.3-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.11manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

fastsafetensors-0.3.3-cp310-cp310-win_amd64.whl (422.9 kB view details)

Uploaded Jul 7, 2026 CPython 3.10Windows x86-64

fastsafetensors-0.3.3-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (1.9 MB view details)

Uploaded Jul 7, 2026 CPython 3.10manylinux: glibc 2.27+ x86-64manylinux: glibc 2.28+ x86-64

fastsafetensors-0.3.3-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl (1.8 MB view details)

Uploaded Jul 7, 2026 CPython 3.10manylinux: glibc 2.26+ ARM64manylinux: glibc 2.28+ ARM64

File details

Details for the file fastsafetensors-0.3.3.tar.gz.

File metadata

Download URL: fastsafetensors-0.3.3.tar.gz
Upload date: Jul 7, 2026
Size: 77.8 kB
Tags: Source
Uploaded using Trusted Publishing? No
Uploaded via: twine/6.2.0 CPython/3.12.12

File hashes

Hashes for fastsafetensors-0.3.3.tar.gz
Algorithm	Hash digest
SHA256	`ba4fb59be8a6adbc91723848c3c6f57a9a9a5d2247d9768f84511380d01e554c`
MD5	`8c953c966fc346db72d7d68331048a39`
BLAKE2b-256	`074cf17bd54c933fd23648ce1e4272adf27d8d7e99b788c8859544bbd39f02e7`

Algorithm	Hash digest
SHA256	`94bd5f003bd9a0d7bc0d262c823ed2ccd74ba0396772a8a327fe1c16f86b6e4c`
MD5	`e70030db4121cf6be8c8ff0606f9efb2`
BLAKE2b-256	`7db86064fd3eeecdb7d8e624ae87773b1a8a8f81111dbbca27aca683e8f9b0c0`

Algorithm	Hash digest
SHA256	`7db5f3ee98433cc8e5dcaaef2517f47c34a51ea8229350406b24b2f84af209d3`
MD5	`c0f3f75a7a1f56865f8614252d5f8960`
BLAKE2b-256	`1a16dc09db973bdc763d5ff64a36acc9e847d0b6c150591d9917ed65f304b722`

Algorithm	Hash digest
SHA256	`e3386834e1595834124cfbb23e05a732cb1b4fb5f98562375c1806f14fe1d677`
MD5	`b38723b1ef20dc93bd5a7d3617a4ac24`
BLAKE2b-256	`2480aeb98bc1575cbd759466568ac6dcbad0a572622cdee0a604fb9604577ecf`

Algorithm	Hash digest
SHA256	`b61a5f6e0e0b0da591f2790d3fbc6eef7f6e8597087977b946069e7bfbb2ec7a`
MD5	`e466880d904e918374313b6d0ae98a3c`
BLAKE2b-256	`3dd34193c0ae305edb94a45f4b0789cf705106b87a1546ca5dafb211c8ab274a`

Algorithm	Hash digest
SHA256	`bbf9fbc5f74d6a3f333777eb222aa63453795dacd7ae7f559316a5398d4942ef`
MD5	`a853dd92baad9d6d0e6df34242b8fc3b`
BLAKE2b-256	`d69dd3db53dd5b5069b44f2b8f6fc5a943f4101178a6f74b48a7fc72dc68f142`

Algorithm	Hash digest
SHA256	`f969b2c818942748ff1bc619e6d575a7a7156a7d488d4c41779fba8c4e1891dc`
MD5	`f2e49237adbe7a40e7480880656a5318`
BLAKE2b-256	`16a2fd30fe6ed5825cb4642bdd102d3c5a32d2d82c91bdb45ae61a9571278cf8`

Algorithm	Hash digest
SHA256	`39c252a5528fa8653366f979d2ab8fa81159db4456b7c72cb5c288adb7699079`
MD5	`8ec440b945036ca4598a8144b99aa1c5`
BLAKE2b-256	`8065388a55e6b2b3023fb732843335de14803f16a6d92f0cd47f1125d28b77ac`

Algorithm	Hash digest
SHA256	`3587bc66b8dec560ad903becf9540889013d4d47f0e10cf45f19bedc7b7bffa7`
MD5	`1c733e3013d51442488caa7547507439`
BLAKE2b-256	`928ce3347b2a44a8ab9aced94fa450df4f309baa21f7f2981a8a7bd6a977f4d3`

Algorithm	Hash digest
SHA256	`2c2f788a936ffd17938484360339645812e14cf1b2bdf4c18c035c713218e5a9`
MD5	`850ca041ea9b4836800ab5044d71214b`
BLAKE2b-256	`d13af95f7fc099ac1fc4c22aa46257d159eac88e29dd0765d21c4fc91caedb01`

Algorithm	Hash digest
SHA256	`be753c669b2e12b744f757303482d95079d48cd7fcce992680b74284b12e6ffa`
MD5	`03567e44f9a9a787f5dffbf2258a500c`
BLAKE2b-256	`1f36e4e2f6bfaac6120e214624ead73ae12e304ad50e5dee7dbb249d9000f01f`

Algorithm	Hash digest
SHA256	`92f8cf8e6617cffe24522023c726c5911c1ca5b7e3249da054abb0fac8d27040`
MD5	`2559af34f8ae7cac25de25e22746f6d1`
BLAKE2b-256	`2ff579f187385b4b9093742fec94e7889d22634ab6eba8403707797246fb2418`

Algorithm	Hash digest
SHA256	`df2f641647b79093c5e3f4bb0b9c7d771f22b61279714c3059d5c71a7cf708da`
MD5	`439430442fc790507d7ebbdf254e72f3`
BLAKE2b-256	`645140bdc05f922251c54518232b2ef47fd925094ee699c93a45c8792628a10e`

Algorithm	Hash digest
SHA256	`2f15b3d293468f2a6aae54b77781521a5d4f0db49bc6cd655442cca2207b3e78`
MD5	`e002fa28aec3af24242e63823122a89c`
BLAKE2b-256	`65e1ed09e65cff92d93bf8b7ec52bd7912395df22b649105d906c5cf34be1c84`

Algorithm	Hash digest
SHA256	`50d8137ce2ca072d1717564c3810d7ccf649eb9f27179f26470258465682bf42`
MD5	`508f3531777c25472d30523736f83224`
BLAKE2b-256	`1428cd0fe7ffe75a84ffe2d7da3e86bdb052f01f4b7442dde44e2837aa34e4ed`

Algorithm	Hash digest
SHA256	`db542d1dfa60dc3c610d9750e3d90107e2ef76feed34958ba8b81814969ea168`
MD5	`13b41b5835d37cc36dc2a4a8fa1652b9`
BLAKE2b-256	`7506dce623c5bc294dc72e4f8eb48d55a5db829b392b3e717a5156226be807b4`

fastsafetensors 0.3.3

fastsafetensors

Performance Highlights

Quick Start

Design Details

Code of Conduct

Development

Publication

Project links

Key dates

1 maintainer

Credits

License

Requires

Provides Extra

Tags

Classifiers

Download files

Source Distribution

Built Distributions

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

Release history Release notifications | RSS feed