NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #2118 most downloaded on PyPI
k-bit optimizers and matrix multiplication routines.
Last release 1 months ago
27 Aug 2026
Release timing varies
gaps range from 2 weeks to 5 months
Most releases are documented
notes for 44 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
4 years old
62 releases · first in 2022
One column per quarter.
ROCm 10.0 build is now included for both Linux and Windows.
Full Changelog: 0.50.1...0.50.2
This release adds support for additional hardware platforms and improves performance on RTX Spark / DGX Spark.
This release adds support for additional hardware platforms and improves performance on RTX Spark / DGX Spark.
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.50.0...0.50.1
This release adds support for additional hardware platforms and improves performance on RTX Spark / DGX Spark.
Full Changelog: 0.50.0...0.50.1
Removed deprecated APIs: the research module, non-blockwise (block_wise=False) optimizers, and legacy dynamic quantization functions, along with their…
This release brings a new fused 4-bit GEMM for inference on CUDA and ROCm, faster CPU ops on x86-64 and ARM64, reduced host-side overhead, and a much improved Apple Silicon backend. We've also added Windows on ARM CPU support, ROCm builds for Windows, additional ROCm and CUDA build variants, and new optimizer support on CPU and Intel XPU.
New fused 4-bit dequantize + GEMM kernels replace the old GEMV and dequantize + F.linear paths for small-to-medium batch sizes. 4-bit inference is up to 4x faster at batch sizes of 2 through 64 across Turing through Blackwell, with wins at batch size 1 in many cases too. Nested (double) quantization and bias are fused in as well, so nested quant sees an additional benefit. Kernel selection happens automatically at runtime based on shape, GPU architecture, and SM count. See #1949 for benchmarks and details.
The SIMT version of the new 4-bit GEMM has been ported to ROCm and wired into the same dispatch for small inference batches, validated on gfx1100, gfx1201, and gfx1151 (#1979).
Stability and performance improvements bring AMD ROCm support out of preview; it is now considered stable.
Apple Silicon support is improved. The MPS backend added optimized Metal kernels from the Hub (#1875) and was substantially improved (#1960, #1983, #1994) so that all 4-bit and LLM.int8() configurations now work on MPS. On macOS 26+, install the kernels package for the best performance, which enables the optimized Metal kernels; otherwise a naive fallback is used. The MPS path requires torch >= 2.9.
The last remaining feature for support parity is the 8bit optimizers, which will land in a future release.
Blockwise quantization and dequantization on CPU are considerably faster, mostly from better SIMD usage plus some compile flag tuning. Improvements range from 1.1x to over 20x depending on op, dtype, and hardware, with the largest gains on fp16 and on x86-64 CPUs without AVX-512. The LLM.int8() matmul on CPU was improved as well. See #1968 for benchmarks.
Reduced Python dispatch overhead, especially on the CUDA/ROCm backend (#1953).
Windows ARM64 CPU wheels are now built with NEON-optimized kernels (#1959), with nightly test coverage added (#1962).
gemv_4bit bf16 correctness on Intel Arc A-series (Alchemist) GPUs (#1942).research module, non-blockwise (block_wise=False) optimizers, and legacy dynamic quantization functions, along with their CUDA/HIP kernels (#1871, #1880).spmm_coo, spmm_coo_very_sparse) and dropped the cusparse/hipsparse dependencies (#1881).igemm, batched_igemm, and check_matmul are deprecated and now emit warnings (#2003).[in_features, out_features] orientation to matmul_4bit now emits a DeprecationWarning. Support is likely to be removed in the future. This is not a typical use case (#1949).This release also includes a number of other improvements, bug fixes, and documentation updates. See the full changelog below.
maxerr1 threshold for fp32 in test_gemv_4bit by @jiqing-feng in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1919gemv_4bit bfloat16 correctness on Intel Arc A-series (Alchemist) GPUs by @jiqing-feng in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1942version when loading CPU gemm_4bit_forward from the Hub by @jiqing-feng in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1972Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.49.2...0.50.0
The default blocksize of 64 for 4bit quantization is now supported on ROCm. Previously the default was 128, which was a mismatch from the default for
Full Changelog: 0.49.1...0.49.2
Update AMD targets by @sstamenk in #1832
Full Changelog: 0.49.0...0.49.1
Remove deprecated code by @matthewdouglas in #1798
CPU performance for 4bit is significantly improved on x86-64, with optimized kernel paths for CPUs that have AVX512 or AVX512BF16 support.
Full Changelog: 0.48.2...0.49.0
Fix indexing overflow issue for blockwise quantization by @matthewdouglas in #1784
Full Changelog: 0.48.1...0.48.2
This release fixes a regression introduced in 0.48.0 related to LLM.int8(). This issue caused poor inference results with pre-quantized checkpoints in
This release fixes a regression introduced in 0.48.0 related to LLM.int8(). This issue caused poor inference results with pre-quantized checkpoints in HF transformers.
Full Changelog: 0.48.0...0.48.1
Fix for warpSize deprecation in ROCm 7.0 by @pnunna93 in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1762
We now officially support Intel GPUs on Linux and Windows! Support is included for all major features (LLM.int8(), QLoRA, 8bit optimizers) with the exception of the paged optimizer feature.
This support includes the following hardware:
A compatible PyTorch version with Intel XPU support is required. The current minimum is PyTorch 2.6.0. It is recommended to use the latest stable release. See Getting Started on Intel GPU for guidance.
We now officially support Intel Gaudi2 and Gaudi3 accelerators. This support includes LLM.int8() and QLoRA with the NF4 data type. At this time optimizers are not implemented.
A compatible PyTorch version with Intel Gaudi support is required. The current minimum is Gaudi v1.21 with PyTorch 2.6.0. It is recommended to use the latest stable release. See the Gaudi software installation guide for guidance.
nn.Parameter by @matthewdouglas in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1720Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.47.0...0.48.0
We now officially support Intel GPUs on Linux and Windows! Support is included for all major features (LLM.int8(), QLoRA, 8bit optimizers) with the exception of the paged optimizer feature.
This support includes the following hardware:
A compatible PyTorch version with Intel XPU support is required. The current minimum is PyTorch 2.6.0. It is recommended to use the latest stable release. See Getting Started on Intel GPU for guidance.
We now officially support Intel Gaudi2 and Gaudi3 accelerators. This support includes LLM.int8() and QLoRA with the NF4 data type. At this time optimizers are not implemented.
A compatible PyTorch version with Intel Gaudi support is required. The current minimum is Gaudi v1.21 with PyTorch 2.6.0. It is recommended to use the latest stable release. See the Gaudi software installation guide for guidance.
nn.Parameter by @matthewdouglas in #1720Full Changelog: 0.47.0...0.48.0
0.48.0: Intel GPU & Gaudi support, CUDA 13, performance improvements, and more!
Compare
Further removal of previously deprecated code
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.46.0...0.47.0
Fix params4bit passing bnb quantized by @mklabunde in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1665
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.46.0...0.46.1
Many APIs that were previously marked as deprecated have now been removed.
torch.compile without graph breaks for LLM.int8().
torch.compile without graph breaks for 4bit.
fullgraph=False.fullgraph=True.torch.library and custom ops APIs. This helps enable our torch.compile and additional hardware compatibility efforts.bitsandbytes.manylinux_2_24 (previously manylinux_2_34).Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.45.4...0.46.0
This is a minor release that affects CPU-only usage of bitsandbytes. The CPU build of the library was inadvertently omitted from the v0.45.4 wheels.
This is a minor release that affects CPU-only usage of bitsandbytes. The CPU build of the library was inadvertently omitted from the v0.45.4 wheels.
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.45.4...0.45.5
This is a minor release that affects CPU-only usage of bitsandbytes. There is one bugfix and improved system compatibility on Linux.
This is a minor release that affects CPU-only usage of bitsandbytes. There is one bugfix and improved system compatibility on Linux.
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.45.3...0.45.4
This is a small patch release containing a few bug fixes.
This is a small patch release containing a few bug fixes.
Additionally, this release contains a CUDA 12.8 build which adds the sm100 and sm120 targets for NVIDIA Blackwell GPUs.
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.45.2...0.45.3
This patch release fixes a compatibility issue with Triton 3.2 in PyTorch 2.6. When importing bitsandbytes without any GPUs visible in an environment
This patch release fixes a compatibility issue with Triton 3.2 in PyTorch 2.6. When importing bitsandbytes without any GPUs visible in an environment with Triton installed, a RuntimeError may be raised:
RuntimeError: 0 active drivers ([]). There should only be one.
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.45.1...0.45.2
This is a patch release containing compatibility fixes.
This is a patch release containing compatibility fixes.
triton>=3.2.0pyproject.tomlpyproject.toml by @SauravMaheshkar in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1373Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.45.0...0.45.1
A number of public API functions have been marked for deprecation and will emit FutureWarning when used. These functions will become unavailable in fu…
PR #1401 brings full LLM.int8() support for NVIDIA Hopper GPUs such as the H100, H200, and H800!
As part of the compatibility enhancements, we've rebuilt much of the LLM.int8() code in order to simplify for future compatibility and maintenance. We no longer use the col32 or architecture-specific tensor layout formats while maintaining backwards compatibility. We additionally bring performance improvements targeted for inference scenarios.
This release includes broad performance improvements for a wide variety of inference scenarios. See this X thread for a detailed explanation.
The improvements were measured using the 🤗optimum-benchmark tool.
For more benchmark results, see benchmarking/README.md.
Example throughput improvement for Qwen 2.5 14B Instruct on RTX 4090:
Example throughput improvement for Qwen 2.5 3B Instruct on T4:
Example throughput improvement for Qwen 2.5 14B Instruct on RTX 4090:
Example throughput improvement for Qwen 2.5 3B Instruct on T4:
The size of our wheel has been reduced by ~43.5% from 122.4 MB to 69.1 MB! This results in an on-disk size decrease from ~396MB to ~224MB.
🤗PEFT users wishing to merge adapters with 8-bit weights will need to upgrade to peft>=0.14.0.
bitsandbytes.functional.int8_vectorwise_dequant(). This functionality is being integrated into 🤗PEFT and 🤗transformers.bitsandbytes.functional module now has an API documentation page.A number of public API functions have been marked for deprecation and will emit FutureWarning when used. These functions will become unavailable in future releases. This should have minimal impact on most end-users.
The k-bit quantization features are deprecated in favor of blockwise quantization. For all optimizers, using block_wise=False is not recommended and support will be removed in a future release.
As part of the refactoring process, we've implemented many new 8bit operations. These operations no longer use specialized data layouts.
The following relevant functions from bitsandbytes.functional are now deprecated :
Additionally the following functions from bitsandbytes.functional are deprecated:
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.44.1...0.45.0
This is a significant release, bringing support for LLM.int8() to NVIDIA Hopper GPUs such as the H100.
As part of the compatibility enhancements, we've rebuilt much of the LLM.int8() code in order to simplify for future compatibility and maintenance. We no longer use the col32 or architecture-specific tensor layout formats while maintaining backwards compatibility. We additionally bring performance improvements targeted for inference scenarios.
This release includes broad performance improvements for a wide variety of inference scenarios. See this X thread for a detailed explanation.
🤗PEFT users wishing to merge adapters with 8-bit weights will need to upgrade to peft>=0.14.0.
FutureWarning when used. These functions will become unavailable in future releases. This should have minimal impact on most end-users.block_wise=False is not recommended and support will be removed in a future release.Fix optimizer support for Python <= 3.9 by @matthewdouglas in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1379
Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.44.0...0.44.1
The AdEMAMix optimizer is a modification to AdamW which proposes tracking two EMAs to better leverage past gradients. This allows for faster convergen
The AdEMAMix optimizer is a modification to AdamW which proposes tracking two EMAs to better leverage past gradients. This allows for faster convergence with less training data and improved resistance to forgetting.
We've implemented 8bit and paged variations: AdEMAMix, AdEMAMix8bit, PagedAdEMAMix, and PagedAdEMAMix8bit. These can be used with a similar API to existing optimizers.
import bitsandbytes as bnb
optimizer = bnb.optim.PagedAdEMAMix8bit(
model.parameters(),
lr=1e-4,
betas=(0.9, 0.999, 0.9999),
alpha=5.0,
eps=1e-8,
weight_decay=1e-2,
alpha=5.0,
)
The block size for all 8-bit optimizers has been reduced from 2048 to 256 in this release. This is a change from the original implementation proposed in the paper which improves accuracy.
A fix to enable CUDA Graphs capture of kernel functions was made in #1330. This allows for performance improvements with inference frameworks like vLLM. Thanks @jeejeelee!
The trend of LLMs to use larger vocabularies continues. The embeddings can take up a significant portion of a quantized model's footprint. We now have an implementation of Embedding4bit and Embedding8bit thanks to @galqiwi!
Example usage:
import torch
import torch.nn as nn
from bitsandbytes.nn import Embedding4bit
fp16_module = nn.Embedding(128, 64)
quantized_module = Embedding4bit(128, 64)
quantized_module.load_state_dict(fp16_module.state_dict())
quantized_module = quantized_module.to(0)
We are now building binary wheels for each change on main. These builds can be used to preview upcoming changes.
:speedboat: Continuous Build
move_to_device kwarg to the optimizer's load_state_dict by @koute in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1344Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.43.3...v0.44.0
The AdEMAMix optimizer is a modification to AdamW which proposes tracking two EMAs to better leverage past gradients. This allows for faster convergence with less training data and improved resistance to forgetting.
We've implemented 8bit and paged variations: AdEMAMix, AdEMAMix8bit, PagedAdEMAMix, and PagedAdEMAMix8bit. These can be used with a similar API to existing optimizers.
FSDP: Enable loading prequantized weights with bf16/fp16/fp32 quant_storage
Params4bit.__new__ post PR #970. It supports models exported with non-default quant_storage, such as this NF4 model with BF16 storage.This release is quite significant as the QLoRA bug fix has big implications for higher seqlen and batch sizes.
This release is quite significant as the QLoRA bug fix has big implications for higher seqlen and batch sizes.
For each sequence (i.e. batch size increase of one) we expect memory savings of:
seqlen=1024, and 4888GB for seqlen=128,00seqlen=1024 and 1258GB for seqlen=128,00This was due to activations being unnecessary for frozen parameters, yet the memory for them was still erroneously allocated due to the now fixed bug.
str2optimizer32bit (#1222, thanks @EtienneDosSantos)Improved the serialization format for 8-bit weights; this change is fully backwards compatible. (#1164, thanks to @younesbelkada for the contributions
…on Linux platforms. This was accomplished by deprecating Make and migrating to Cmake, as well as implementing new corresponding workflows. Huge thanks…
pip install bitsandbytesv0.42 to v0.43, when using 4bit quantization, models may generate slightly different outputs (approximately up to the 2nd decimal place) due to a fix in the code. For anyone interested in the details, see this comment.This release is made possible thanks to the many active contributors that submitted PRs and many others who contributed to discussions, reviews, and testing. Your efforts greatly enhance the library's quality and user experience. It's truly inspiring to work with such a dedicated and competent group of volunteers and professionals!
We give a special thanks to @TimDettmers for managing to find a little bit of time for valuable consultations on critical topics, despite preparing for and touring the states applying for professor positions. We wish him the utmost success!
We also extend our gratitude to the broader community for your continued support, feedback, and engagement, which play a crucial role in driving the library's development forward.
This release added 4-bit serialization, implemented by @poedator, to bitsandbytes. With this,you can call model.save() and model.load() for models tha
This release added 4-bit serialization, implemented by @poedator, to bitsandbytes. With this,you can call model.save() and model.load() for models that contain 4-bit bitsandbytes layers meaning you can save and load 4-bit models. All of this is integrated with the Hugging Face transformers stack. The 0.42.0 release also comes with many bug fixes. See below for detailed change logs.
bitsandbytes.__version__ @rasbt #710bnb.nn.Embedding @neel04 #563Nothing published for this version
Nothing published for this version
Fixed an issue where 4-bit serialization would fail for layers without double quantization #868. Thank you, @poedator
Bug fixes:
Nothing published for this version
Nothing published for this version
4-bit serialization now supported. This enables 4-bit load/store. Thank you @poedator #753
Feature:
Fixed bugs in dynamic exponent data type creation. Thank you @RossM, @KohakuBlueleaf, @ArrowM #659 #227 #262 #152
Bug fixes:
Release 0.41.0 features an overhaul of the CUDA_SETUP routine. We trust PyTorch to find the proper CUDA binaries and use those. If you use a CUDA vers
Release 0.41.0 features an overhaul of the CUDA_SETUP routine. We trust PyTorch to find the proper CUDA binaries and use those. If you use a CUDA version that differs from PyTorch, you can now control the binary that is loaded for bitsandbytes by setting the BNB_CUDA_VERSION variable. See the custom CUDA guide for more information.
Besides that, this release features a wide range of bug fixes, CUDA 11.8 support for Ada and Hopper GPUs, and an update for 4-bit inference performance.
Previous 4-bit inference kernels were optimized for RTX 4090 and Ampere A40 GPUs, but the performance was poor for A100 GPUs, which are common. In this release, A100 performance is slightly improved (40%) and is not faster than 16-bit inference, while RTX 4090 and A40 is slightly lower (10% lower).
This leads to approximate speedups compared to 16-bit (BF16) of roughly:
Features:
Bug fixes:
Documentation:
User experience:
Performance:
Deprecated:
Fixed a but where a non-existent LD_LIBRARY_PATH variable led to a failure in python -m bitsandbytes #588
Bug fixes:
Nothing published for this version
Added precompiled CUDA 11.8 binaries to support H100 GPUs without compilation #571
Features:
Bug fixes:
Documentation:
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
This release adds efficient inference routines for batch size 1. Expected speedups vs 16-bit precision (fp16/bf16) for matrix multiplications with inn
This release adds efficient inference routines for batch size 1. Expected speedups vs 16-bit precision (fp16/bf16) for matrix multiplications with inner product dimension of at least 4096 (LLaMA 7B) is:
The inference kernels for batch size 1 are about 8x faster than 4-bit training kernel for QLoRA. This means you can take advantage the new kernels by separating a multi-batch 4-bit query into multiple requests with batch size 1.
No code changes are needed to take advantage of the new kernels as long as a batch size of 1 is used.
Big thanks to @crowsonkb, @Birch-san, and @sekstini for some beta testing and helping to debug some early errors.
Features:
Bug fixes:
device variable for bitsandbytes layers to be compatible with PyTorch layers.Deprecated:
pip install bitsandbytes and need to be compiled from source.This release brings 4-bit quantization support for QLoRA fine-tuning and a critical bugfix that doubled the memory cost of 8-bit models when they were
This release brings 4-bit quantization support for QLoRA fine-tuning and a critical bugfix that doubled the memory cost of 8-bit models when they were serialized. Furthermore, paged optimizers are introduced, including 8-bit Lion.
Features:
Bug fixes:
Deprecated:
4-bit matrix multiplication for Float4 and NormalFloat4 data types.
Features:
Bug fixes:
Deprecated:
Added Fake FP8 layers for research purposes (available under bnb.research.nn. ...)
Features:
bnb.research.nn. ...)Nothing published for this version
Nothing published for this version
In this release, CUDA 10.2 and GTX 700/K10 GPUs are deprecated in order to allow for broad support of bfloat16 in release 0.39.0.
This release brings 8-bit Lion to bitsandbytes. Compared to standard 32-bit Adam, it is 8x more memory efficient.
Furthermore, now models can now be serialized in 8-bit and pushed to the HuggingFace Hub. This means you can also load them from the hub in 8-bit, making big models much easier to download and load into CPU memory.
To use this feature, you need the newest transformer release (this will likely be integrated into the HF transformer release tomorrow).
In this release, CUDA 10.2 and GTX 700/K10 GPUs are deprecated in order to allow for broad support of bfloat16 in release 0.39.0.
Features:
python -m bitsandbytes now gives extensive debugging details to debug CUDA setup failures.Bug fixes:
Improvements:
Deprecated:
Nothing published for this version
Nothing published for this version
This release changed the default bitsandbytets matrix multiplication (bnb.matmul) to now support memory efficient backward by default. Additionally, m
This release changed the default bitsandbytets matrix multiplication (bnb.matmul) to now support memory efficient backward by default. Additionally, matrix multiplication with 8-bit weights is supported for all GPUs.
During backdrop, the Int8 weights are converted back to a row-major layout through an inverse index. The general matmul for all GPUs by using Int8 weights is done by casting the weights from Int8 to the inputs data type (FT32/FP32/BF16/F16) and then doing standard matrix multiplication. As such, the matrix multiplication during backdrop and for non-tensor-core devices will be memory efficient, but slow.
These contributions were the work of Alexander Borzunov and Yozh, thank you!
Nothing published for this version
The 0.36.0 release brings a lot of bug fixes, improvements, and new features:
The 0.36.0 release brings a lot of bug fixes, improvements, and new features:
Additional features also include fake k-bit quantization and smaller block sizes for block-wise quantization, which are used in our k-bit Inference Scaling Laws work. Fake k-bit quantization is useful to simulated k-bit data types, but they do not provide memory or runtime benefits. Here is how you use these features.
Faster block-wise quantization that now allows for very small block sizes of down to 64:
from bitsandbytes import functional as F
q, state = F.quantize_blockwise(X, blocksize=64)
X = F.dequantize_blockwise(q, state, blocksize=64)
k-bit fake quantization via block-wise quantization:
# 4-bit float quantization stored as Int8
from bitsandbytes import functional as F
# 4-bit float with 2 exponent bits
code = F.create_fp8_map(signed=True, exponent_bits=2, precision_bits=1, total_bits=4).cuda()
q, state = F.quantize_blockwise(X, code=code) # q has 4-bit indices which represent values in the codebook
X = F.dequantize_blockwise(q, state)
Features:
Regression:
Bug fixes:
Improvements:
Fixed a bug in the CUDA Setup failed with the cuda runtime was found, but not the cuda library.
Bug fixes:
Fixed a bug in the CUDA Setup which led to an incomprehensible error if no GPU was detected.
Bug fixes:
Fixed a bug where the CUDA setup failed due to a wrong function call.
Bug fixes:
CUDA 11.8 support added and binaries added to the PyPI release.
Features:
Bug fixes:
This release introduces memory-efficient backprop through frozen weights where the gradient is calculated from the 8-bit weights but is computed in fp
This release introduces memory-efficient backprop through frozen weights where the gradient is calculated from the 8-bit weights but is computed in fp16. This is useful for creating Low-rank (LoRa) Adapters for fine-tuning large models.
This is a feature contributed by @dbaranchuk and @justheuristic.
Features:
memory_efficient_backward=True which enables backprop of gradients through frozen weights.Bug fixes:
CPU quantization now supports a variable blocksize variable to enhance quantization speed or precision. 19a7adca7a6c9bf7061a384d7e9d9b13676a1a88
Features:
blocksize variable to enhance quantization speed or precision. 19a7adca7a6c9bf7061a384d7e9d9b13676a1a88Bug fixes:
We thank @mryab, @mbrukman, @chessgecko, @dbaranchuk for pull request with bug fixes and new features.
Nothing published for this version
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →