Hacker GitHub

← All repos

microsoft/BitNet

View on GitHub ↗ ★ 41,444 C++

Official inference framework for 1-bit LLMs

22 mentions on Hacker News by 16 people: 5 stories, 17 comments · first 2024-10-18 · latest 2026-08-23

Mentions per year 22 mentions · busiest year 2026 (11)
20082026
AllStoriesComments NewestTopOldest
  1. …coming for sure. Model on a custom silicon: https://chatjimmy.ai/ 1-bit models that run on a CPU: https://github.com/microsoft/BitNet
    comment by ForHackernews · 2026-08-23 · in Fable and the end of the free lunch
  2. …Microsoft BitNet - Official inference framework for 1-bit LLMs https://github.com/microsoft/bitnet Ternary Mirror-symmetrical Arithmetic And Its Application To Digital Signal Processing by Alexey Stakhov (related: Nikolay
    comment by peter_d_sherman · 2026-08-20 · in AVX Bitwise ternary logic [VPTERNLOGD] instruction busted (2024)
  3. …data centres are left holding the bag. I suspect Apple ends up laughing all the way to the bank. https://github.com/microsoft/BitNet
    comment by ForHackernews · 2026-07-23 · in Alphabet's cash burn raises alarm for Big Tech as AI spending climbs
  4. …of running self-hosted LLMs will be reduced by an order of magnitude once this hits mainstream. [1] https://github.com/microsoft/BitNet
    comment by josteink · 2026-05-11 · in Local AI needs to be the norm
  5. …there is a new model, they recently released a new inference engine for CPUs for bitnet models: https://github.com/microsoft/BitNet/blob/main/src/README.md
    comment by naasking · 2026-03-12 · in BitNet: Inference framework for 1-bit LLMs
  6. ▲ 2
    Microsoft BitNet – inference with 1.58 bits
    story by Flere-Imsaho · 2026-03-11 · 1 comment
  7. …number of threads. Up to 73 tokens/sec with 8 threads (64 toks/s for their recommended Q6 quant): https://github.com/microsoft/BitNet/blob/main/src/README.md
    comment by naasking · 2026-03-11 · in BitNet: Inference framework for 1-bit LLMs
  8. It comes from (intentionally?) misleading docs: https://github.com/microsoft/BitNet/issues/391 (only suggesting that it's intentional because it's been there so long)
    comment by webXL · 2026-03-11 · in BitNet: Inference framework for 1-bit LLMs
  9. ▲ 370
    BitNet: Inference framework for 1-bit LLMs
    story by redm · 2026-03-11 · 167 comments
  10. ▲ 2
    Show HN: 26x speedup on BitNet sparse ops with AVX-512 and 2-bit encoding
    story by HyperFoldUK · 2026-01-13 · 0 comments
  11. …https://github.com/ikawrakow/ik_llama.cpp and their 4Bit-quants? Or maybe even Microsofts Bitnet? https://github.com/microsoft/BitNet https://github.com/ikawrakow/ik_llama.cpp/pull/337 https://huggingface.co/microsoft/bitnet-b1.58-2B-4T-ggu…
    comment by LargoLasskhyfv · 2026-01-07 · in A 30B Qwen model walks into a Raspberry Pi and runs in real time
  12. …will help immensely with running LLMs locally without having to spend sheer $$ on high end GPU's. https://github.com/microsoft/BitNet?tab=readme-ov-file
    comment by syedhtanveer · 2025-07-22 · in Why Don't You Just Use GPT or Claude?
  13. ▲ 2
    BitNet 1.58bit GPU Inference Kernel
    story by galeos · 2025-05-20 · 0 comments
  14. Repo with demo video and benchmark: https://github.com/microsoft/BitNet "...It matches the full-precision (i.e., FP16 or BF16) Transformer LLM with the same model size and training tokens in terms of both perplexity and end-task…
    comment by hu3 · 2025-04-16 · in Microsoft researchers developed a hyper-efficient AI model that can run on CPUs
  15. https://github.com/microsoft/BitNet
    comment by dkrajews · 2024-11-20 · in 1-Bit AI Infrastructure
  16. and the repo for this project: https://github.com/microsoft/BitNet
    comment by skavi · 2024-11-20 · in 1-Bit AI Infrastructure
  17. …Bitnet into the discussion and explain how its performance relates to this announcement, if at all? https://github.com/microsoft/BitNet "bitnet.cpp achieves speedups of 1.37x to 5.07x on ARM CPUs, with larger models experiencing greater…
    comment by andrewstuart · 2024-10-25 · in Cerebras Inference now 3x faster: Llama3.1-70B breaks 2,100 tokens/s
  18. https://github.com/microsoft/BitNet "The first release of bitnet.cpp is to support inference on CPUs. bitnet.cpp achieves speedups of 1.37x to 5.07x on ARM CPUs, with larger models experiencing greater performance gains. Additionally, it…
    comment by andrewstuart · 2024-10-19 · in AI engineers claim new algorithm reduces AI power consumption by 95%
  19. Here is the Microsoft implementation: https://github.com/microsoft/BitNet
    comment by andrewstuart · 2024-10-19 · in AI engineers claim new algorithm reduces AI power consumption by 95%
  20. Here’s the Microsoft Bitnet implementation. https://github.com/microsoft/BitNet From the Readme: “The first release of bitnet.cpp is to support inference on CPUs. bitnet.cpp achieves speedups of 1.37x to 5.07x on ARM CPUs, with larger…
    comment by andrewstuart · 2024-10-19 · in Ask HN: What are the implications of Bitnet for Nvidia,the economy and AI?
  21. The energy claims up to ~70% can be verified. The inference implementation is here: https://github.com/microsoft/BitNet
    comment by Randor · 2024-10-19 · in AI engineers claim new algorithm reduces AI power consumption by 95%
  22. ▲ 173
    Microsoft BitNet: inference framework for 1-bit LLMs
    story by galeos · 2024-10-18 · 33 comments