{"data":{"slug":"bitsandbytes-foundation-bitsandbytes","name":"bitsandbytes","tagline":"Large language model quantization toolkit for PyTorch.","github_url":"https://github.com/bitsandbytes-foundation/bitsandbytes","owner":"bitsandbytes-foundation","repo":"bitsandbytes","owner_avatar_url":"https://avatars.githubusercontent.com/u/175231607?v=4","primary_language":"Python","stars":8385,"forks":900,"topics":["llm","machine-learning","pytorch","qlora","quantization"],"archived":false,"github_pushed_at":"2026-07-29T18:27:51+00:00","maintenance_label":"Very active","url":"https://www.graphcanon.com/tools/bitsandbytes-foundation-bitsandbytes","markdown_url":"https://www.graphcanon.com/tools/bitsandbytes-foundation-bitsandbytes.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/bitsandbytes-foundation-bitsandbytes","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=bitsandbytes-foundation-bitsandbytes","description":"Accessible large language models via k-bit quantization for PyTorch.","homepage_url":"https://huggingface.co/docs/bitsandbytes/main/en/index","license":"MIT","open_issues":54,"watchers":52,"ai_summary":"bitsandbytes library enables k-bit quantization in PyTorch for making large language models more accessible. Supports various hardware platforms and accelerators including NVIDIA, AMD GPUs, Intel XPU, and CPUs.","readme_excerpt":"## System Requirements\nbitsandbytes has the following minimum requirements for all platforms:\n\n* Python 3.10+\n* [PyTorch](https://pytorch.org/get-started/locally/) 2.4+\n * _Note: While we aim to provide wide backwards compatibility, we recommend using the latest version of PyTorch for the best experience._\n\n#### Accelerator support:\n\nNote: this table reflects the status of the current development branch. For the latest stable release, see the\n[document in the 0.50.0 tag](https://github.com/bitsandbytes-foundation/bitsandbytes/blob/0.50.0/README.md#accelerator-support).\n\n\n##### Legend:\nšŸš§ = Planned |\n怰ļø = Partially Supported |\nāœ… = Supported |\nāŒ = Not Supported\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
PlatformAcceleratorHardware RequirementsLLM.int8()QLoRA 4-bit8-bit Optimizers
šŸ§ Linux, glibc >= 2.24
x86-64ā—»ļø CPUMinimum: AVX2
Optimized: AVX512F, AVX512BF16		āœ…āœ…āœ…
šŸŸ© NVIDIA GPU
cuda		SM60+ minimum
SM75+ recommended		āœ…āœ…āœ…
šŸŸ„ AMD GPU
cuda		\n CDNA: gfx908, gfx90a, gfx942, gfx950, gfx1250
\n RDNA: gfx103X, gfx110X, gfx115X, gfx120X\n 		āœ…āœ…āœ…
šŸŸ¦ Intel GPU
xpu		\n Data Center GPU Max Series
\n Arc A-Series (Alchemist)
\n Arc B-Series (Battlemage)\n 		āœ…āœ…āœ…
šŸŸŖ Intel Gaudi
hpu		Gaudi2, Gaudi3āœ…怰ļøāŒ
aarch64ā—»ļø CPUāœ… *āœ…āœ…
šŸŸ© NVIDIA GPU
cuda		SM75+āœ…āœ…āœ…
šŸŖŸ Windows 11 / Windows Server 2022+
x86-64ā—»ļø CPUAVX2āœ…āœ…āœ…
šŸŸ© NVIDIA GPU
cuda		SM60+ minimum
SM75+ recommended		āœ…āœ…āœ…
šŸŸ„ AMD GPU
cuda		\n RDNA: gfx103X, gfx110X, gfx115X, gfx120X\n āœ…āœ…āœ…
šŸŸ¦ Intel GPU
xpu		\n Arc A-Series (Alchemist)
\n Arc B-Series (Battlemage)\n 		āœ…āœ…āœ…
arm64ā—»ļø CPUāœ…āœ…āœ…
šŸŽ macOS 14+
arm64ā—»ļø CPUApple M1+āœ… *āœ…āœ…
ā¬œ Metal
mps		Apple M1+āœ… *āœ…šŸš§
\n* While supported, these marked features may lack in performance optimizations.\n\n---\n\n## License\n`bitsandbytes` is MIT licensed.","github_created_at":"2021-06-04T00:10:34+00:00","created_at":"2026-07-11T23:31:59.846629+00:00","updated_at":"2026-08-04T00:01:48.760254+00:00","categories":[{"slug":"inference-serving","name":"Inference & Serving","url":"https://www.graphcanon.com/categories/inference-serving","markdown_url":"https://www.graphcanon.com/categories/inference-serving.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/inference-serving"},{"slug":"llm-frameworks","name":"LLM Frameworks","url":"https://www.graphcanon.com/categories/llm-frameworks","markdown_url":"https://www.graphcanon.com/categories/llm-frameworks.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/llm-frameworks"}],"tags":[{"slug":"llm","name":"llm"},{"slug":"machine-learning","name":"machine-learning"},{"slug":"pytorch","name":"pytorch"},{"slug":"qlora","name":"qlora"},{"slug":"quantization","name":"quantization"}],"trust":{"provenance":{"is_fork":false,"github_id":373674258,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-08-04T00:01:47.941Z","maintenance":{"label":"Very active","score":96,"methodology":"github_public_v1","releases_90d":2,"days_since_push":5,"last_release_at":"2026-07-29T18:27:56Z"},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-11T23:32:03.824Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-08-04T00:01:48.407Z"},"languages":{"value":["python"],"source":"github.language+pyproject.toml","observed_at":"2026-08-04T00:01:48.407Z"},"license_spdx":{"value":"MIT","source":"github.license","observed_at":"2026-08-04T00:01:48.407Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["When you need advanced k-bit quantization on PyTorch for hardware like NVIDIA GPUs with SM75+ recommended.","For projects that aim to leverage AMD RDNA GPUs like gfx1201 under Linux or Windows."],"when_not_to_use":["Avoid if your setup includes Intel Gaudi processors as QLoRA 4-bit support is partial and 8-bit optimizers are not available.","Steer clear if you require full compatibility with ARM-based CPUs, as specific GPU optimizations might lack coverage."],"source":"enrich:decision_facts","observed_at":"2026-07-17T05:22:12.343Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"bitsandbytes provides k-bit quantization in PyTorch, enhancing large language model accessibility across multiple hardware platforms."}]}}