Visitar URL original
Release 2026-03 follow-up · Issue #2026 · winpython/winpython · GitHub
Skip to content

Release 2026-03 follow-up #2026

Closed
Closed
Release 2026-03 follow-up#2026
Milestone
2026-03 towards AI local / 3.15.0rc / jupyterlite

Description

@stonebig
stonebig
opened on May 13, 2026

date: mid-august (free-threading push)

wanted:

  • reducing the number of objects per release by 25%: no more"whl" version
  • wppm usable in jupyterlite and xeus: from wppm import piptree;pip = piptree.PipData();print(pip.down('.'))
  • modernize infrastructure to be compatible with Python-3.16 and Trusted Publishing
  • numba-0.66.0, but so no numpy-2.5.0 yet
  • using "paid" CLAUDE Code to simplify maintenance and improve quality
  • ai local alpha but getting clother, via https://github.com/microsoft/onnxruntime-genai, but Phi-3-mini can't be included ... but Google chrome does download its own model... so a question of when-not-if
  • no Python-3.15.0rc1 [60% of] slim with available binary packages... not the practical level of 80%

remark:

  • python-3.15 free-threading should get massively rapid adoption, because of speed-up, abi3.abi3t, 50% of users, 80% of dev being AI
  • ... example: scaling free-threading 32x (vs 16x) on numpy+cpython-3.15 three months before release : 30x faster vs 5x via multi-processing
  • for further speed, bits of rust will be incorporated in cpython around 3.17: "Python interface, Rust engine"
  • jupyterlite / pyodide is getting more flexible, because of PEP-783 allowing pypi wheels
  • scikit-learn starting to support gpu acceleration
  • new Windows "Python Install manager" is progress
  • python at 50% of pypl.github.io (tutorial search) in august 2026: guess is 30% of humans, 70% of AIs ... so why not 60% at christmas

Activity

  1. stonebig
    added this to the 2026-03 AI local / 3.15 free-threading with binary packages milestone on May 13, 2026
  2. stonebig
    pinned this issue on May 14, 2026

  3. stonebig commented on May 14, 2026

    @stonebig
    stonebig
    on May 14, 2026
    ContributorAuthor

    audit-trail:

    • wppm -p generated tree has 10 unecessary blanks at start of each line
    • wppm -r generated tree has 10 unecessary blanks at start of each line
    • remove un-needed noise on constraints search like wppm -r pandas![.]
    • reduce indentation of wppm tree to 4, to match python default code indentation
    • signaling functional imperfection to pipdetree on required extras (both should have the same definition of truth)
    • suggesting pyodide/micropip to accept pylock.toml files (as a normalized "pip freeze")
    • switch default sourceforge download to 3.13.12.0slim (cycle 2026-01)
    • local build housekeeping: remove local pylock, requirement after sucessfull check
    • WinPython 2026-03 cycle creation: rename "free" flavor as "dotf", more easy to understand
    • pandas-3.0.3
    • numpy-2.4.6
    • wppm for xeus: don't use pip._vendor, only packaging
    • "whl" variant has dowloaders....
    • piptree ref from packaging.markers import Marker not working in initial cloud build.... wppm for cloud: don't use pip._vendor, unless no packaging package yet
    • wppm-17.5.20260524-py3-none-any.whl
    • wppm usable in jupyterlite and xeus-jupyterlite: from wppm import piptree;pip = piptree.PipData();print(pip.down('.'))
    • wppm-17.5.20260525-py3-none-any.whl: back to usable also in cloud build
    • do we want to do universal requirements ? (for amd/arm) https://www.peterdebelak.com/blog/generating-a-fully-qualified-and-hashed-requirements-file/
    • lxml event: abi3t free-threading-wheel ? Not yet
    • scikit_learn-1.9.0
    • github billing system appears: seems still free if you don't use AI.

    b0 (push for wppm and pipdeptree working on jupyterlite, non-published)

    b1 (2026-06-14)

    • remove "whl" specific from local and cloud build. having a local wheelhouse doesn't make sense now that pip manages pylock.toml
    • SIMPLIFY: in local build, cleanup local files for wheelhouse hashes (if not just removing whl builds)
    • SIMPLIFY: in cloud build, replaces hashes.md per a variable name
    • wppm-17.6.20260621 : show missing dependancies (from wppm vs pipdeptree comparison)
    • remark: the two most prolific cpython developers now also use (and sign with) claude
    • python-3.15.0b3
    • cython-3.2.6, keras-3.15.0, altair-6.2.2
    • cvxpy-1.9.1, dask-2026.6.0 (compat with pandas-3.1) , polars-1.42.0
    • spyder-6.1.5
    • it appears, as much Qwen popular is, it has a bias, like mistral can have, or openai.

    b2 (2026-06-28)

    • remark: Qualcomm to Acquire Modular, june 24th 2026
    • remark: pixi.dev being a successor to mamba and uv ? https://github.com/prefix-dev/pixi
    • experimenting with python install manager (microsoft store , or winget install 9NQ7512CXL7T )
      • wppm -md works
      • sqlite_bro works
    • remove sv-ttk (be back to minimal, as tcl9 is progress)
    • SIMPLIFY: in local build, the method to get a python core is now modernized to "Python Install Manager"
    • SIMPLIFY: learning curve to use AI Claude, via an initial teething experience on simpler sqlite-bro
    • SPEED-UP with Claude Fable5: wppm-17.6.20260715 now 5x faster on worst case
    • SIMPLIFY with Claude Fable5: pypi wppm page now based on a dedicated README_PYPI.md
    • SIMPLIFY with Claude Fable5: wppm-17.7.20260715 now only reports cycles once instead of 8 times
    • TRUSTED PUBLISHING of wppm-17.7.20260715
    • PRE-AI BUG FIX with Claude Fable5: wppm-17.7.20260716 now "wppm -l pandas[.] -l0 gives one line per extra, and all extras again
    • RE-PURPOSING wppm-17.8.20260718 as JSON able and Python-Distribution agnostic, so usefull for bigger Python community
      • --json output mode for options -p -r -ls -md
      • removal of WinPython reference when it's not a WinPython environnement
    • Continuing MY JOURNEY TO Use, claude opus: sqlite-bro-1.3.2
    • Continuing MY JOURNEY TO Un-tying WPPM from WinPython and Windows: wppm-17.9.20260805
    • python-3.15.0rc1, python-3.14.7 (python-3.13.15 not yet published par astral)
    • numba-0.66.0 , but not compatible with numpy-2.5
    • remark: Numba-0.67.0 may land its first support for windows-on-arm, couldn't happen before LLVM was on windows on ARM
    • converging on a first ai-local: onnxruntime-genai + hugginface-hub (model phy from microsoft not included: it's 1.5 Gb or more)
    • streamlit_bokeh-3.9.1, bokeh-3.9.1
    • SIMPLIFY: local Python binary build extracted from "py"
    • jupyter-builder-1.0.2, jupyter-server-2.20.0, jupyterlab-4.6.1, notebook-7.6.0
    • REMOVING: astropy, distributed, django
    • re-include opencv (40Mo), so exclude geopandas eco-system
    • onnxruntime_genai
    • compress -mx7 to stay below 650Mo
    • mssql_python-1.13.0 gave back 10 Mo
    • python-3.13.15
    • remark: only numpy-2.6 will support python-3.15, so python-3.15 slim is out of this cycle window

    b3 (2026-08-08)

    • IMPROVE: wppm more compatible with venv, as venv is the trend nowodays
    • remark: Claude Code doesn't necessarly see what you don't tell it, like "piptree must keep working on jupyterlite"
    • REMOVE pypy support: pypy gave many speed idea to cpython, with the utf-8 strings still to come
    • REMOVE pre-requirement local build step
    • ADD wppm CI tests, as it would prevent old bug found per Claude (also Claude Opus 5 only finds randomly bugs, if not hunting)
    • on MY JOURNEY TO Use Claude agentic: good ideas, then a proposed good idea that is bad in the context
    • REMOVE pydub: un-maintained / security issue
    • MAINTAINABILITY: build is with python >=3.11, wppm is ok with python >=3.10
    • numpy-2.5.2 is here for python-3.15.0rc1 \o/ !
    • python-3.15.0: cffi, pillow
    • cosmetic fix on local build logging: a little 💖 or ⚠️ at end of local log files
    • python-3.15.0 slim, as numpy as a wheel for it on pypi
    • python-3.15.0 slimf: 270 packages (vs 469 for 3.14 slimf, 297 for 3.15 slim)
    • numba-0.67 + numpy-2.5.2 (scaling better on free-threading)
    • openai-3.0.0 (preparing for astra overload ?)
    • 3.14 slimf slightly clother to parity : adding qdldl, shapely... 90% (missing qt stack, optimizers, statsmodels, opencv)
    • RE-ENGINEERING with AI: prepare a more re-usable CLOUD workflow for next cycles , with build_winpython_cycle.yml
    • RE-ENGINEERING with AI: do a simpler LOCAL workflow, with winpython_builds.toml
    • FEATURE: implement a wppm --top-level that allows to filter a minimal requirements from a distro or requirements.txt
    • wppm-17.11.20260813
    • on MY JOURNEY TO Use Claude agentic: beware of addiction, as indeed the model starts to think like you after a while...
    • a dependabot security alert: need pydantic-ai-slim>=1.106.0 .... updating the workflow
    • remark using Claude AI: attention is [still] all you need

    rc (2026-08-1x)

    • scipy-1.8.1
    • still no matplotlib for 3.15 free-threading... so no point of publishing any 3.15 slim
    • python-3.15.0 next-cycle wait of wheels is:
      • pythran <--> scipy --> Scikit-learn (to get a reasonnably not so slim 3.15)
      • statsmodels -> plotnine (statsmodels intended to go free-threading with 3.15)
      • pyarrow (october 1st) --> streamlit + duckdb
      • pyyaml is a big blocker
      • markupsafe is a big blocker
      • jupyter stack
      • having more stable-abi wheels would help ... more
    • SIMPLIFY: wppm.piptree easier to use in jupyterlite ?
    • adjust further Distribution size ?
    • an opencv torch ultralytic thing ?

  4. stonebig commented on Jun 14, 2026

    @stonebig
    stonebig
    on Jun 14, 2026
    ContributorAuthor

    WinPython 2026-03 b1 (2026-06-14)

    Changes from WinPython 2026-02 :

    • Infrastructure:
      • python-3.13.14
      • python-3.14.6
      • python-3.15.0b2 (warning: broken tcl/Tk-9 till b3)
      • wppm package can now be installed on jupyterlite and jupyterlite-xeus
    • New Packages:
    • Upgraded Packages:
      • matplotlib-3.11.0, polars-1.41.2, pip-26.1.2, pipdeptree-3.1.0
      • scikit_learn-1.9.0, streamlit-1.58.0, panel-1.9.3
    • Removed Packages:
    • Differences among builds:
      • slimf (python-3.14 free-threading) has no Qt Stack and is missing still lot of binary packages

    Areas of particular interest for testers:

    • pipdeptree-3.1.0

    Built via Github Actions, where binaries are kept 66 days:

    Binaries at:

    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    65131be135a9bf090dbe26e7ff86ae39 2b7cac4a2d82c71ee03cbe7ec339e719bceab4d3 4b82c83e2ffb1654e608d96322c9aacb3f986f9965075e1307220d4073f996df WinPython64-3.13.14.0dotb1.exe 17 049 926 Bytes 7545b038b5306359975549d81efd4bd6893a1a889cb2f7112958f37984e6e9d8
    46930098dc6ed4befbeefc15f513946a 8d0dbc183ca3fadeb6faf67ff2a95b762026f9ec 3c8d6514b8c5736ff03ab2bc9223f460c1e224ef90e6e6db33894f9edf5bc98c WinPython64-3.13.14.0dotb1.zip 27 760 481 Bytes 487ba41f5ef0c91ff1ff07447778fb82f9fd770f081656efe53fa5e22c39a8af
    567a46ed5a8a58da12b9aa8004195b1a 0d4daf876e2383cd7978a903bfa4556ce513757c 28c6501d6d6c12352e2c0f2c718f2778bb213dcca00856301eb2e164ad688b20 WinPython64-3.13.14.0slimb1.7z 674 173 554 Bytes d749f03cb971161a657973edba3f58f39a470dbb128045180d0246bf2bad4c5c
    63e4f4633760cd66782ccd814d8c2741 651e0ac0b0434b5ca33fed058282439a75b24c0e 8297e42d9f85ac0efd9cda91c0add2420bfd659f09949e38fa53213b68e1f2de WinPython64-3.13.14.0slimb1.exe 674 387 873 Bytes 90a04db56d8aa64ca28e8718de228167fa1ccd76caeff807a89af1db68b3b96a
    2e02c0bbadb1909a9c30a4817e88ebdc 5832b69a08696a6f2312afd713d6c2b3eab4574e af1ce308692439442f033350961b2bbf0997eb80d403c51d9fc1b1ef7ac066a8 WinPython64-3.13.14.0whlb1.7z 690 040 249 Bytes 6e602ee1badd2446e78ea487e1e216f54be6b95ac340d6fd80256f98af4d4f1c
    0ac0c4c61cfcdfce1bea9b9e4c551e42 be2351daa1edbec550d323b5ee17b551b5342824 24d3dd741a6ac0a5853f2dca9f2bac811d3c6c3b19b7b9eaa93d8457f5d2c8b8 WinPython64-3.14.6.0dotb1.exe 17 504 915 Bytes f8edaa16db783c6c5f35a2ce8177d0ea5e7fef98c5077e82ee45da434926786b
    75171b672e6f29a9e5ebc5904d396c0b afc8c2c6686ec3edc1dc8c40ee1a3fb18bb0b031 cf900d392677794aa39b345ee209be414399046c8fba2bc7ada08bd972a049bd WinPython64-3.14.6.0dotb1.zip 28 365 814 Bytes 0fd7cb972e39de060bd7598d595c27d048dd5226e81c0c866487e54d9fb21a98
    c84e2c430e5dd4338bf7fcbfa5ce5a70 ccb290f223a8ef158b1d3c4ca037ea1e99c56d00 547a0cc34ee9938fda89edc39079ee2bdf09a2dd3e8c8ac290ce30f781890192 WinPython64-3.14.6.0dotfb1.exe 17 585 882 Bytes 2636a01c0cb2c8059bd9873391cd024376aa48287626bfeffbcb2f01bd0198dd
    0cecb02c9a5da7564ddff9d9d5f333ab 32ae76685a29dce25b6fd8dcbf37e9a5677615bf e19dc8ad0f7f95a61d542cb55f338b486986c805c25fcd10920b073663406d91 WinPython64-3.14.6.0dotfb1.zip 28 569 305 Bytes 7a316cc7c5de0a3f164fc8535100787a0a6c7b68110e59ef2ebfe4000fb20798
    7bc6816a74e00ced962394678ffce8aa 8007c5c4b972e9b1f1ced4360e749806b640e5c3 03a6cafa8c9640bab0148325cb235a011ef045ff93512e9450b1d6412af13f26 WinPython64-3.14.6.0slimb1.7z 679 905 096 Bytes c41a30068a6e4f1e45767c06a1c20a664b3948172050d2369f0e96353d16c013
    f6eb955799cbcdda641f039247a5c6c6 3b92d7f176526a85bad3056e145e318cce467147 9ec48d61e1bde4c041290be85a53f20749efe6bc14864387c01df23540f7a3a5 WinPython64-3.14.6.0slimb1.exe 680 119 146 Bytes 3ad7c4129a83686e1ffd540a6abfef60ae8746ce0c06447a8fb9a971f0fb679c
    8a0554adaca6cf5e82101e38c1321cbb be1f9a167a2535207207a0ccaf3e36313184f915 48d07505e32204fcbb12c0fa6779e733710ad8b3c96c648882311e4ab3a80eaf WinPython64-3.14.6.0slimfb1.7z 459 602 276 Bytes d5498ca67ff4347a5d7b6eb55bc00ef0fb6947e2d4a1a6650da5eadab93575bd
    59e61458ac12ce26c433c18a0b8e137a 154f4a1b04881bd6f855490250fe6d8b5f9788c9 6a1add0acfe1308d8499fd4c013169404ac0c04d36f82dbff89aab48913cd209 WinPython64-3.14.6.0slimfb1.exe 459 816 296 Bytes 3ae0442dc1a6d8969bb87a9557cef0a4db55dcc0a339e472bdd088531cc08017
    8538d5db1a81c3db559a56ff40e97928 a2e935d94e909d55e578008f9b2bff18a9043a15 1bc179089737a293be694d4543ed0ff14283ea6b8a1d9631801fcf6331e85a82 WinPython64-3.14.6.0whlb1.7z 695 327 913 Bytes 7f14ace87a2b42ff7c54cb2642b8510a1772ca75f3c511e012e698f6d38050c3
    204e9363566c8dc4f551ecdac7367ab9 f7b069d990450a0124399d6fae9f61b19d890f4c 70b73d51cb2317eabd015e499b5d40b936e17743e5fb4cebb5890a08428182c0 WinPython64-3.15.0.3dotb1.exe 17 720 221 Bytes ef87cd53072e60b01e6acce9c3bc875cf09afc3efe15e8e5832982b68b3252de
    c6924713af747b0d9c11c6243499ead3 423077c4faac59e9999b8e396a86dd42a0e38a95 e31edd2fdf4a1f014ec9e3524fbbce1251e48f59430a815e474c95dc8b2dd358 WinPython64-3.15.0.3dotb1.zip 28 037 830 Bytes 79caff01d75c4e98afa9f876f2085e8ddb1283626956739ac7d61c6f2baf5b8e
    228b5bfc6a1186dc9d57005c8fe62040 493306d5de1c1b7834c7670bea6655d7f21bd156 7fea0432ee5f05cdeb256490ee54afde8d45ebea32d8b8d6f7512664a9b67879 WinPython64-3.15.0.3dotfb1.exe 17 737 065 Bytes c7d8763b005f92a42539c29dcea72d99e37e30556ff3d15baca039a765ad24bd
    31fdfbc0e43390b3ea251e314438303a cc7e3fac5213a50881b62eaadcf773d13a036f06 58f04035e15fcc2c59a748f8f20070331ceba020c4e5d20a9d632b60f9ba3321 WinPython64-3.15.0.3dotfb1.zip 28 113 277 Bytes cbf645a291b87085e857be36ec4f190d72337cf6f3ea4bf4ca8527d5737448ba

    Associated lockfiles, and typical usage with pip:
    pip install --no-deps --require-hashes -r pylock.64-3_13_14_0slim.toml --no-warn-script-location

    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_13_14_0dotb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d
    ed07c7266afc8e8c0f3d67a4c5e10f12 3e67f7b74ae34ff62ffad390ec0b31ca79412881 27b7122be44d192ec61857dd6d72e6c4b7fdb8d97c5d6994b2ede6917e0c5fae pylock.64-3_13_14_0slimb1.toml 202 128 Bytes afd55d84b9df629e381efa29231c0d12e87725f14592456cd5dc65c2037c52ec
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_13_14_0whlb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d
    0e58ac64c6b5b2ef12ff15ce9c74b11f 2e603dc555e5441bb223282df3df2de298f5bacb 84ce1f0bd208c60e1edd5ebb066e4ecb5d74f53b1c7067ba75726bf4c1910186 pylock.64-3_13_14_0whlb1_wheels.toml 192 459 Bytes 6654682736872904f73efc80ea42ac9deb2176bf40cb122014845b0e030bc13e
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_14_6_0dotb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_14_6_0dotfb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d
    9d648f79de431a6107b8cebb8e124ef4 c6e8abb22dc99e18bf6648a26e36fccfd0a24db0 dd53e001342e3e966f9cb6fe258cb1ee0468e4ace5d4b51ae3258c114a7f0580 pylock.64-3_14_6_0slimb1.toml 202 092 Bytes b27ec8f0b1b6144fd5acd9518b96750d64a5e38b2d23d3fc02161210a8338821
    916c2af304b301f649606672e2c68d6a 201560382be3404d6258916bbeed6d559c4458db c6fede5b42c218d5d0bcadccfba67c6e725de4d7ffcfe78fddc61043f813eb88 pylock.64-3_14_6_0slimfb1.toml 182 766 Bytes 78f99c9a1a469e7353d76052b3ea7f5eab0fa7e13e676bb469697b1b78c1410a
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_14_6_0whlb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d
    2261f1f66b4aae50aec8de786f0c05cd 652fc679eb7b21fe8ed3db6efabac601419aa7ef 6c869b192d3b65dc476cba022efc085467b69a8316b0ed400b36290e512d7721 pylock.64-3_14_6_0whlb1_wheels.toml 192 423 Bytes a44029d30bfc8bf88d7741d594556d7d64d949232a327b362617978e095f27c2
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_15_0_3dotb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d
    cfac3e6b51e6c629dd28f211bc1814fa f072f5cf9345164bc17c608277174292b01e9962 a0cc66ab03e94583dca1c5051d90d71663d88098fa41e99420bd4e5ea40ea5af pylock.64-3_15_0_3dotfb1.toml 3 428 Bytes 79329bea0de720a0d9c562c8943b6e11aef7f8c1e625a6f43d801b4b5119e54d

  5. stonebig commented on Jun 28, 2026

    @stonebig
    stonebig
    on Jun 28, 2026
    ContributorAuthor

    WinPython 2026-03 b2 (2026-06-14)

    Changes from WinPython 2026-03 b1 :

    • Infrastructure:
      • python-3.15.0b3 ( repaired tcl/Tk-9 till b3)
      • wppm package shows missing dependancies, for example :
    wppm -p pandas![test]
pandas[test]==3.0.3 ,
    pytest-xdist==? >=3.6.1;extra==test
    • New Packages:
    • Upgraded Packages:
      • altair-6.2.2, duckdb-1.5.4, polars-1.42.0, scipy-1.18.0
      • spyder-6.1.5
    • Removed Packages:
    • Differences among builds:
      • slimf (python-3.14 free-threading) has no Qt Stack and is missing still lot of binary packages

    Areas of particular interest for testers:

    • pipdeptree-3.1.0

    Built via Github Actions, where binaries are kept 66 days:

    Markdowns are posted in the comit tree like : https://github.com/winpython/winpython/blob/672dce022ef413b8a9caad5772794c732167ada8/changelogs/WinPythonslim-64bit-3.13.14.0.md

    Binaries at:

    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    1dcccf3dab6c6670327bfe44ed66ecce 0d2b9b4e2e26d76e24387329f535ac1b7f458959 55c7cf715f734f7e68ef269115757ec0547962c21511e28ecc94c26cc3cec128 WinPython64-3.13.14.0dotb2.exe 17 048 628 Bytes d7eef36c2f9334a25432834dce8197f5243d97ebbcf2b44c8acc0b06fffb6418
    0d00c8c2da3c5d89b9afe649612d0a58 e76ccdf6ec54c1aeb9e865822ec65dbddca5449d aa7b580eff61f8a2251586bb180713f57ce819f5060d16d54545d4c3705f2a06 WinPython64-3.13.14.0dotb2.zip 28 072 262 Bytes b9d0f43ae1c71a2d0ee538212a16078a23906d3c3356e7161949cd5032a35aeb
    1d2f027e8fabd29099ee55211ccec691 6cb61326a2b08ce97f87da3acab2e82be4a5ad70 9fd9da423e492fb57b797d9f92d0df9359b7f0535d35990943cdf6ee83f2b194 WinPython64-3.13.14.0slimb2.7z 677 188 654 Bytes e8af704d28fc2af85227bf4fb53cb0abf2aeadd21c22f0181f4153708c36780b
    5213c6904a7444389ff412604d84abb6 f6aab6200a02e5f54ff825dcac51df3bcb9eb781 745b370c004f95e4650fcdc37927cf0bbefc37d50a3a11f2512a5985a85ffcbc WinPython64-3.13.14.0slimb2.exe 677 405 084 Bytes f1d805b7889aa87b58fa73e08ec3755ac8fe369aff5786f6e06cb1f6f6e812d3
    4cb84c12c9e82a0bb2481d7d06fa2d74 6f9c5402de5d823e3a630a212b16c0606fb794b7 1544a71221376a616266bb43ae31d131e3dfd78d43f0cff71a895116616de906 WinPython64-3.13.14.0whlb2.7z 691 521 877 Bytes dc5b337e8bc8a8b30bf3e4e01db95c9beb9c4d83608a7ab9edb92b818c27a3b7
    56b912a650f8c17a519f3250a7b8faea 07e158e5a7b359aff7f3b200d71b727e26b21a2c 6464763f02c5c13294a8c88601f7b7920e9c955f930f6fac6128ba001676da3e WinPython64-3.14.6.0dotb2.exe 17 502 984 Bytes e5885f0ed8ca49dcfbd18cfbddeafb636ea417fdccfea5e4992bdd01cea645b8
    b550bcac266c118260b6b948319d89f1 332b54b02dda7ec7880f12f1e4c4a0d317620f0c 322086aaf84095f3f947ccfe8d349603ec51bde78c1d939a4331f9e40f7783a7 WinPython64-3.14.6.0dotb2.zip 28 705 583 Bytes 2243e93bab740b9acaaaad9602821e1d1a8f5d938b4aefd8c9669ca586fe18cf
    40b6e2442ff12bbc70eba1a740b4c974 ee664acdd13a81997c1a3825d66612310594c2dd 323778887f46e0a04f9d480e1f44e13725515de7774ee91ba8a408cd22cb6a11 WinPython64-3.14.6.0dotfb2.exe 17 588 648 Bytes ccbf01c1d2c51f5d0fc3579db9e16d60fccdf6bd5306ac28fc404cfacbb02c6e
    7109ada93ab1e5d8b10f186093e8c094 d7d77b4237ede85cfdcb4edf2810681ab3eb8a6e 0a00b30c9685df4dd0f71e3f8f7af85f3b13d8c6c6351843f49d232ce56df31a WinPython64-3.14.6.0dotfb2.zip 28 939 030 Bytes 160b31e031a3a558e6ee4ab8cc7133d11fa366b3000d638c19ca4b85be4c99d6
    246fbb65aead8aa8741ab2e78156db8b 8c9b29a085531851f72926dff34cbf1af8c2baed c2722f8e92f37cda0f4aa8bf1c8b6e9862ee487ad41f179e93f349dbcd25bd2f WinPython64-3.14.6.0slimb2.7z 683 426 485 Bytes fb0d21e6207cbce0261e62031101e14a4e86d167c70f78060ce661bcee8f8af6
    5ee6b7277b48d53ce574e838f1e6f000 9f551a9c768d27710a8a0fac9da13c660dbf0a64 2c47e21d7848db5f6074ef774395c4eb2aba87f99ad626fa6b2a3698e8cb5658 WinPython64-3.14.6.0slimb2.exe 683 641 798 Bytes 35028532dbeeec8941f2817a11b7dcde3a2ebe4d564041e64f22314c308f00f2
    cea050c641a93ebca68a97feed6ab306 b059e7d93ecc9540dbe117a1f3adbb6b91b07b4c cad79b0e2807b22cac29f0ccfb32966af9f327cf61a65fef7cbb8c40985a0386 WinPython64-3.14.6.0slimfb2.7z 461 115 829 Bytes c02058f4f2b6079c7ad35085047a262a77837ff374dca13d33ebb560d5948b2c
    f6ad0d89832ddbbb9c6a9be24ec01e2b 1bd8617763ad6f84a25104c9906cb27ef5120283 35d51024b49368a93902b9d330db6b9878ff8a97f6ceed2724d3c3895b7561b5 WinPython64-3.14.6.0slimfb2.exe 461 330 348 Bytes 0b3d53a061e5baf2a5ecf9e0f6dfd93cd37f8ab590babfb0cbac7cba103fd276
    34397be57bbe0977aae1087da440b83f caa4a78cc6482e47d18eab74aaa6b25865aed48f 64139b28f547ae33a6ccb74a609d2c4251adf29b3e30746a5c5aeab7f93e71b9 WinPython64-3.14.6.0whlb2.7z 697 515 879 Bytes d909f16ea1e555eb858203276024a0ea9dd0f24b5fdfbd7bb1294ee84456989b
    b7e6664c102fe8049682cdae52226695 d7a3f9d84eb0be7ac6d48f573dfe3b9598ba158e 1dd3ae26d9016ad16fd0a737bfb499c3d16882291767a722da96d48a8fff8177 WinPython64-3.15.0.3dotb2.exe 19 481 558 Bytes 42e7f06ea0aa5e982e09e81c0ebe8ef65d7f9e7d81c2332702ab4542238177e1
    ac7b003bc72d69cceedf58bedcc57a58 b925eba92776dd24687fb4765ff71550ec11095a ffd7650b746660851d75401ac26110844640d123284da1dff15f0c3628df8ac5 WinPython64-3.15.0.3dotb2.zip 30 254 150 Bytes 7c2eb56e608e67eed2c139c900c7ebf19a878b8198bb8b494d7305b4ba73127a
    2e46ee5d656e8eb5eafb861a59d283ad 697f82f0d5f40cdde3e3ebd447cbaa865eb498ec aa2ade78bd34fce709d9c621c72a11d70a61a76e053f02e05d34a1a84d68abf2 WinPython64-3.15.0.3dotfb2.exe 19 497 846 Bytes 59515a965e100c81e1f85c086303e8ce2aef57af0b40e9d6cbfe731525874cd2
    0282d286b0331e35e741a544c5fbb625 71c70de6f9c5e3ed8aff88173188c78e235b97cb bac291b58e69b8a8a325756975cf3922d9ed3a008d1b742554e2ebf32ec60301 WinPython64-3.15.0.3dotfb2.zip 30 336 144 Bytes da9b1ef14dbb4bf91f0c302dac5bb3a7b39a87f92f27ae4cda567fbdce19bdb8

    Associated lockfiles:

    typical usage with pip:
    pip install --no-deps --require-hashes -r pylock.64-3_14_6_0slimb2.toml --no-warn-script-location

    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_13_14_0dotb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c
    f2aec23ce7b3c80bc20c7793cf40c05d a41469bf206a66aebf9b9db742831ba459eb0266 d507fbb3dd7d5801f7f3a645b1cfdc15e8d36b3134117493bd826374b7b199d8 pylock.64-3_13_14_0slimb2.toml 202 953 Bytes f31916267113763f48d636f512cc62d664036bce81e4e837545998aa4bd83e75
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_13_14_0whlb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c
    c8317768b112548c93e178974f5cc342 41e4077385eb11f3e406591aed10e3fe253babe3 2dbb5eed38ba430d7cae2bb5d16a06dddbe8ad3b54eaf0318771bf817fbbf53e pylock.64-3_13_14_0whlb2_wheels.toml 193 284 Bytes a45ed0bc0d3e80d002a35420903875b2cf63efccf4427d70d9e5f3f1a84f59ce
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_14_6_0dotb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_14_6_0dotfb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c
    c0fa83f33fadf034cf9cd287cadbcb8e af2076e37d4954e983c3235091b2a2d5f3c018d8 1c823b953553b240aae1232164fec11226bf43d194acb3b80b1270a611961b2c pylock.64-3_14_6_0slimb2.toml 202 917 Bytes 0f8ce00bdc874840895162553d8da93512f3e9ae08ae8dc03e477ecce62375b2
    d271e9dd29fec36df1193a853c52555a feabbddf29f61b656ae6f8eaf1390fcd35c0dafc af75215bf900d1ff5ae792ddbcd6e5e9d10d6867202642342be7730d7e94fbb3 pylock.64-3_14_6_0slimfb2.toml 182 784 Bytes 2e3fd4eff446a23bcf71ee02a4a5114b34fc895107261df76fa434495613c05f
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_14_6_0whlb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c
    1e8a2e9ada75ae56b5892aef7e7bb764 fe29d129b4e56e6d9192e032a639e8407d354f2b 8cf3258ee16468aeca6762ada726a5171c2b6523feaf385e652a3955dac1e566 pylock.64-3_14_6_0whlb2_wheels.toml 193 248 Bytes df0649359c78d82c5d69ad3ec714b5328774d58ff4711bb21e4b7bab0e8c6789
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_15_0_3dotb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c
    1dd11e35d5991e4857c472ec87b0b766 90ff959d0b290ffa37ce9acb64596910b5568dbb d7af6131f01b82ec300c4fb0185340f4ae1cb8c4c8191e926f1533091ed9f741 pylock.64-3_15_0_3dotfb2.toml 3 428 Bytes 49459f1ba68bfc3a4a3d14196f9f1b637a9934e0eed373b6e0c727f481e38a6c

  6. stonebig commented on Aug 8, 2026

    @stonebig
    stonebig
    on Aug 8, 2026
    ContributorAuthor

    WinPython 2026-03 b3 (2026-08-08)

    Changes from WinPython 2026-03 b2:

    • Infrastructure:
      • python-3.13.15, python-3.14.7, python-3.15.0rc1
      • wppm and sqlite-bro "trusted publishing", with Claude AI help
      • wppm has now a dedicated readme: https://github.com/winpython/winpython/blob/master/README_PYPI.md
      • increasing compression to -mx7 to keep below 650 Mo
      • pip-26.2.1
      • with pip-26.2 now handling pylock.toml, the .whl is loosing relevance
    • New Packages:
      • opencv-python, onnxruntime-genai, mssql-python-odbc
    • Upgraded Packages:
      • cython-3.2.9, mypy-2.3.0, onnxruntime-1.28.0, duckdb-1.5.5, polars-1.43.2,
      • spyder-6.1.6, pandas-3.0.5, matplotlib-3.11.1
    • Removed Packages:
      • pymongo, geopandas, pyogrio, pyproj, astropy, django
    • Differences among builds:
      • slimf (python-3.14 free-threading) has no Qt Stack and is missing still lot of binary packages

    Areas of particular interest for testers:

    • stability while a first wave of rework bone by Claude
    • python-3.15.0rc
    • will publish the pylock.toml of the slim version on the web page

    Built via Github Actions, where binaries are kept 66 days:

    Binaries at :

    Next build: rc

    • tweaking for size
    • reworking WinPython Welcome Page : links to pylock files
    • with pylock.toml manage per pip, maybe the whl version is to replace per a pylock.toml link ?
    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    58bafd23db68f374b0dc00783201cf16 26d83bf7d71e7225f91463d206af67f49456d483 9f94a7951da53769081bc9ca472de2db3873ffc613c62bc0677550437c0468ca WinPython64-3.13.15.0dotb3.exe 16 904 968 Bytes b3222e60f7114d47ff46618c8180634162659a5f0332719d90f497c94d2611b1
    c531ba64325bbe31ae31cfb4f759e0cf d84769d9f2b1e8a895f859b2f250502d672ae447 cef685b79f8c146252524f64835d4dfd13f14d2243bf7c6f08df2a2178db824a WinPython64-3.13.15.0dotb3.zip 28 141 464 Bytes a4cffd81f13a02e672d0900b623769d2f3c5a10a33109d40d7626590587f4ed8
    3e80a926f1a7b243b10e0ab596f70df2 aed8112dbb4bf65388a026b04338bbebc3fc3893 ccd51ab8a206fb2ccd5b86f1e139900a68db3ac425a34ccb6486f84cf7ebeaf8 WinPython64-3.13.15.0slimb3.7z 661 079 534 Bytes 4fcb75b30adcefaa16170b91a57e717e4aab6bc995993674a9c4476e5ac4c1b2
    e6fce4807d38c7d08c166956d3503a38 adb3fc29688a03f89238b642c9819e9092a99831 030d8250833cca7e8823f8be03452a10b5ad324a6d1ea2448a32dd0319db9ec7 WinPython64-3.13.15.0slimb3.exe 661 294 839 Bytes def38fb7a8fafe2841cbbea87b3d2fb26a3a1c2b24ead8f63d79215a16c7b909
    8218a4f061bef2166f937bd54b39a1f6 9c73ad0f9dffbe855e310591cd20edff75772f0a c719627a097707e1b4bb96d4eb40caef55a2a94526e0d16cfbd839fdebdc944a WinPython64-3.14.7.0dotb3.exe 18 022 665 Bytes a649e650a10f989360caba94d557d7ddc7fca57433373447987999829f7426b9
    92918d83fc4f467a86b79edaac87b1f3 375e512984b7386262e0ac278753b20591a3d12e 83943d189a3a3355a68ffbf1db0fb5b85b99b0ecd5926c9f6ad9adb378f757ee WinPython64-3.14.7.0dotb3.zip 28 646 209 Bytes 9826ebdca2b6abd4c75fc63aaa432d8bde29859d744fe17f6313b2be1ca995f7
    ff287b3fc2ea32c4d435a007c383bc17 ec4ac9eeab2f7ee063177c2d1e1b4c46ed4b0f13 42aec27486077d812da815a3136185051c01f8374c3eba20029926127f456e40 WinPython64-3.14.7.0dotfb3.exe 18 102 286 Bytes 2d1048a1c0ea1ec94028251f56d26faac33615296e43f438b46833e865bffece
    1069cdbd005b4432bb1f0fc57d02ec0c 1b1b2d36110785d9def0a5f268a7b0e51031035a 68d7bfab8aad0ef79b5073f7b6f774b063f058e78b02042be42828e45623a450 WinPython64-3.14.7.0dotfb3.zip 28 872 966 Bytes 4aaeffa613749ae523f0e4a5d5b53458042120a6804e57f3571d521aadeae386
    73ffaf00db8de35d1fbd7ebc892522e5 9b26ea6e4cb62213bf95f1d3c55166f5e520be4c c107b0a8dff7d4687a3eaefaf8d415e2b032500aa43703fe99928a0ca381fb78 WinPython64-3.14.7.0slimb3.7z 666 105 253 Bytes caa700c42b1c940d723cc73637b7660b19d5c1a519fe2f36175f2c61bd212d25
    b7e755a7dc7c60afa5d6067be2339814 8dcfa0c784c2b22be6c185257e38fce751f5a8d6 ab9a20ac6ac02d0b9ef78a4f441f9d138f5909e60ff4f07d8fad2a81f2a69339 WinPython64-3.14.7.0slimb3.exe 666 321 235 Bytes 5c3ded5db99c45a1a14a3b5d3eb41d3eda2672325cc51ea8524dd1b32ee8c685
    8b61208adde3b44d961ce09f6d5755e6 d91ac6999e0e47897eb366c6b46a9e0f9daf1269 6cc869a4e61193c9331f74b8d7cc4a1f3c5005609529cb68cde3139305e41d5f WinPython64-3.14.7.0slimfb3.7z 454 157 006 Bytes f2aa0c85c3d97079c0450ed6f8b7f7b727e6e492601fe3d5b3be180aed0bfaf9
    9688bdb42304afb9f93c1585b1d5b226 3da8f6d179899ea21d6682f8e76ad5a5e685a888 f786558d35a45faafe4aab9faf065d50ef26b162740d67a3a758dd00099033f1 WinPython64-3.14.7.0slimfb3.exe 454 372 683 Bytes a90236998c97a7aa41d24935a0ea4cf4c88bd1282d1f066395d9923f571a3d98
    9c7c42f7cbac054b12c178c5d9729389 5952a82e7df6d0dd06ee99d43f5df1ef4d8b6a64 aae19a750894db58aef961c0620b3a0ffc08fa0685613c4992413f4af2817782 WinPython64-3.15.0.4dotb3.exe 18 832 502 Bytes 48ca4f074dff4f149fc2c112a78807aa0b40ab97fc0668aeb85d5adfa0d444f6
    c854ac5bc0f7564b2154fdb3cc6fff5b 86d12587f0eca3a6c84d9af1014630c164826722 d835c9a3c9e42e3aa1b0a747170e44365b4f7c931123baa820b4d350aec79cd4 WinPython64-3.15.0.4dotb3.zip 29 854 732 Bytes 970ffad06e79e26d5b0924ff0ccc60ed42b04c40b4639bcb6432f196012a7d20
    00189fa14323e70ddb75911e8adcccd9 2b22e0ae2e49b17f2d3fcb1f24826f4128065aa8 00045e84c1466e4136c729be6a8f1a5f40081d0578be042f2f0a70e1d2af743b WinPython64-3.15.0.4dotfb3.exe 18 850 286 Bytes 985b5e8d3b0f21a47e044cd5b5a27008265c3e5037c25e9c6b6b424164443267
    67e94a95c59da24bb23414e4b0660102 e8bf8dc28ec2ec02ac3caabfb04d20807c42c4b4 8816f30cc053def2c8066b27d8f8801a0577f8f2cc8cdbe5289c6b88a61bcd0e WinPython64-3.15.0.4dotfb3.zip 29 932 997 Bytes ecd03b51ee01f6c96978ebdc40108130e436f795b9f7f999b60f185da523a7e8

    Alternative requirements-with-hash256 and lockfiles, and how-to use with pip:
    . with a requirement file:
    . pip download --dest C:\your_temp_Wheelhouse --no-deps --require-hashes -r requir.64-3_14_7_0slimb3.txt
    . pip install --no-deps --no-index --trusted-host=None --find-links=C:\your_temp_Wheelhouse --require-hashes -r requir.64-3_14_7_0slimb3.txt
    . with a pylock file:
    . pip install --no-deps --require-hashes -r pylock.64-3_14_7_0slimb3.toml --no-warn-script-location

    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    a0509239f66c07f75369c1585e75f640 e5e1225ed6bfcf00266e9ef8c3b57749a2d1d5c5 9cfe445de8d4b16655cdba967081c1741b92762629bf66ed13ef9193139fe4cc pylock.64-3_13_15_0dotb3.toml 3 059 Bytes 9d653a70f8cbc37ad24b5d3c2b7281795fb88faff305413e33b50f6b3ed979b6
    52364ca5a92e8ec9f8c99d7f56dd7afe d8513a9aa06979d8b5f1e10a841e2e008c2b5f8c 3764cd651d6321dfb624d79288d2b88626db243c8791f810c8bbca0176da05c4 pylock.64-3_13_15_0slimb3.toml 198 879 Bytes d8b9b627d455a4aa55aaf4e544a113854b882b179df55f9c3cb49c5c68bf22b7
    a0509239f66c07f75369c1585e75f640 e5e1225ed6bfcf00266e9ef8c3b57749a2d1d5c5 9cfe445de8d4b16655cdba967081c1741b92762629bf66ed13ef9193139fe4cc pylock.64-3_14_7_0dotb3.toml 3 059 Bytes 9d653a70f8cbc37ad24b5d3c2b7281795fb88faff305413e33b50f6b3ed979b6
    a0509239f66c07f75369c1585e75f640 e5e1225ed6bfcf00266e9ef8c3b57749a2d1d5c5 9cfe445de8d4b16655cdba967081c1741b92762629bf66ed13ef9193139fe4cc pylock.64-3_14_7_0dotfb3.toml 3 059 Bytes 9d653a70f8cbc37ad24b5d3c2b7281795fb88faff305413e33b50f6b3ed979b6
    a3252b973860552f7b009d93d161f9f5 16b5f56c720f68c83a0336b0cc8f7431d96ff159 f49e8c0b1ae7ff4b5163495ffe2c71a643b6ceca0bf5828665abcd094858c709 pylock.64-3_14_7_0slimb3.toml 198 861 Bytes 5e3af3af36d922a56a1c50a99a23034a0723b68d04e9f3492cf52dd35d2a0296
    b44b1b1c2153e689e7a604ef5718fcdd d3dfa2b1609b6ee05f4e2c9dafc5aa013c7775a3 5e2b0d15e7c2671b0db095dc45c3fe1d2e880b0f92cba7756db3adfa1a260a9a pylock.64-3_14_7_0slimfb3.toml 182 831 Bytes 89c33bec862b4b5028a1a162cc50cb48bea90b2788e95df13a0931d057fb420b
    a0509239f66c07f75369c1585e75f640 e5e1225ed6bfcf00266e9ef8c3b57749a2d1d5c5 9cfe445de8d4b16655cdba967081c1741b92762629bf66ed13ef9193139fe4cc pylock.64-3_15_0_4dotb3.toml 3 059 Bytes 9d653a70f8cbc37ad24b5d3c2b7281795fb88faff305413e33b50f6b3ed979b6
    a0509239f66c07f75369c1585e75f640 e5e1225ed6bfcf00266e9ef8c3b57749a2d1d5c5 9cfe445de8d4b16655cdba967081c1741b92762629bf66ed13ef9193139fe4cc pylock.64-3_15_0_4dotfb3.toml 3 059 Bytes 9d653a70f8cbc37ad24b5d3c2b7281795fb88faff305413e33b50f6b3ed979b6
    3ec2c16c58318d73e0c6957966198fda d2a4645e6e0be6772b49bb11582f96d4bed17db9 33dd2085acbed8807d40220ce599d9f8c6b4ed1aee324bcb6369eb1d3678231a requir.64-3_13_15_0dotb3.txt 835 Bytes 704e36023ab5202622fa6cedd2d20d838f28a73acea13e07d2f523fc393db6df
    e27a731d9ba1d09020529afedb41e82d 1ef017263faa2b2885505dae7419b9e58efe2e18 33112c99fda1ab2366731b4431fb14381e8ecb0f2fe2ddd8d038ea2df8f16969 requir.64-3_13_15_0slimb3.txt 54 593 Bytes 10cdaabc2c53f4bb466e951ba7bb44cf5f23501e5b37b044eca63af72d6b04dd
    3ec2c16c58318d73e0c6957966198fda d2a4645e6e0be6772b49bb11582f96d4bed17db9 33dd2085acbed8807d40220ce599d9f8c6b4ed1aee324bcb6369eb1d3678231a requir.64-3_14_7_0dotb3.txt 835 Bytes 704e36023ab5202622fa6cedd2d20d838f28a73acea13e07d2f523fc393db6df
    3ec2c16c58318d73e0c6957966198fda d2a4645e6e0be6772b49bb11582f96d4bed17db9 33dd2085acbed8807d40220ce599d9f8c6b4ed1aee324bcb6369eb1d3678231a requir.64-3_14_7_0dotfb3.txt 835 Bytes 704e36023ab5202622fa6cedd2d20d838f28a73acea13e07d2f523fc393db6df
    7d6a7fba87287d181156ce49e9ee6eee 56d61c4b46ec0ba3813b4941903f4d56bfd4160c f9c7b6ee274c2dc0f030b8d7cb8de5bce2d70ad11bd99c48a9b4ac9cb4b76ac8 requir.64-3_14_7_0slimb3.txt 54 593 Bytes e39a917b4f04ae822ec5ff6fba8692cb41048ab3cea01ded29a1b5747d11a46f
    83c497dcf0f23478bb88f26548efcd24 c6d14a3af3529f60049ae83c1b82d320bd32cdf3 e0b561f945e7582b423345fb5fb0c6a255d06299502da9e193254132dd275166 requir.64-3_14_7_0slimfb3.txt 50 256 Bytes efba3c1e24276c1c4e4cb2e567a78738fe07c88bc507b0ea4ccc55a9a3564738
    3ec2c16c58318d73e0c6957966198fda d2a4645e6e0be6772b49bb11582f96d4bed17db9 33dd2085acbed8807d40220ce599d9f8c6b4ed1aee324bcb6369eb1d3678231a requir.64-3_15_0_4dotb3.txt 835 Bytes 704e36023ab5202622fa6cedd2d20d838f28a73acea13e07d2f523fc393db6df
    3ec2c16c58318d73e0c6957966198fda d2a4645e6e0be6772b49bb11582f96d4bed17db9 33dd2085acbed8807d40220ce599d9f8c6b4ed1aee324bcb6369eb1d3678231a requir.64-3_15_0_4dotfb3.txt 835 Bytes 704e36023ab5202622fa6cedd2d20d838f28a73acea13e07d2f523fc393db6df
  7. stonebig
    removed this from the 2026-03 AI local / 3.15 free-threading with binary packages / jupyterlite milestone on Aug 8, 2026

  8. stonebig commented on Aug 13, 2026

    @stonebig
    stonebig
    on Aug 13, 2026
    ContributorAuthor

    WinPython 2026-03 (2026-08-22)

    Changes from WinPython 2026-03 b3:

    • Infrastructure:
      • preparation of a permanent "build_winpython_cycle.yml" for next cycle
      • simplification on local build workflow
      • wppm-17.12.20260816 update, with "--top-level" helping to :
        • clean-up a requirements.txt file: wppm -tl requirement.txt
        • summarize the origin of a package need : wppm -tl -v
        • limit the package tree walk-through: wppm -tl -p
      • python-3.15.0rc1 slim is only 60% of a standard slim, so not worth publication
    • New Packages:
      • from pydantic-ai-slim-1.106.0 update (security fix)
    • Upgraded Packages:
      • wppm 17.11.20260816
      • pydantic_ai_slim-1.106.0, numba-0.67.0, numpy-2.5.2, mypy-2.3.0
      • jupyterlab-4.6.2, openai-3.0.0
    • Removed Packages:
    • Differences among builds:
      • slimf (python free-threading) have no Qt Stack and is missing still lot of binary packages

    Built via Github Actions, where binaries are kept 66 days:

    Binaries at:

    next cycle:

    • Python-3.15.0 final should reach 90% of packages availability
    • Usefull Local AI pursuite
    • WinPython build further simplifications
    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    68c57d2bd5fdc15f247d700b97ba691c ba6aa7e194b70af020cb178cca55b28c0d76afaa 4b62ce4f5a6ccee239f9b025a626d3ddf606916fad8814758dcaf2b7aadf4dea WinPython64-3.13.15.0dot.exe 16 915 861 Bytes 9798edc523abb91ce754566f92298f9fa4193442b186fc78bdea56b927b63dc0
    d878b461e29ad52bfba46859d23267dd fa75f45cb0586ac6ba5aa2cd749d8e5207d2e2ea 28e36408f0140c50b207ea059a599c664564e68a3cbb835f03a71f4601efd8f1 WinPython64-3.13.15.0dot.zip 28 158 845 Bytes 4bd77e36e304580c889283cb4c4409c083aa7740e2057ebe440c2b8fe9da0c0e
    f3ef6cdcc1ae1d2d8b913a727192964e e7bcaa3f7a8c31c7382d50b5fd76d21be5a75e47 bcb8a5c951d89f87c69b0660585a4e5ed14d932d3f552632621abdcc0a1f649d WinPython64-3.13.15.0slim.7z 666 143 896 Bytes db22745b2e9990c690894dd5f0c970cf50d5b0279980fa864f7cc89017a6c3b6
    a1292c2500436742868b907810700fb5 79618e3acb2badbe85e784cf6965c3edde43f6a8 774e2a4ddc1c21f36353039c8c2dfbf50ba3779ff08cc1952d5ef5fd9658e655 WinPython64-3.13.15.0slim.exe 666 359 286 Bytes ca8b95b72ce5c620345faceeba5a1e2c5af959c523c53c7b27463cc79ae6554b
    4f5c0159a8ad05ef48d4e2fa7537ffa6 a2c1af53b4f5fc54a460dc2922959367750d99cb 7b461ea5147b7d3be74a879b8ecf18292940142f2df5d3e268e482ecf7d32f24 WinPython64-3.14.7.0dot.exe 18 033 648 Bytes de3216ea8a8842b0967d95e59591d31f30b1df6c94001f309601835654431595
    be677d5d47ed9f79b963986dd3ad8bb4 091717aade3d3f548581edeb17d7692ad42a33a1 dbabedfb50eeb3c2c63dc43c9cb6239eae4a582c3bfd9a5f2ffd00a09b49a527 WinPython64-3.14.7.0dot.zip 28 663 262 Bytes 114e628aa489175addc64319b24aab280b3151c5a4d4a3324fc5eae1f283e049
    8468583680a5409439d2e4dfd2bb8b01 6f668a19963c4a6151d032e01253bb27a4219c01 a929b1a3adba698b9e0713c5689d9bd5483e52f4356fc847d3075acbb94d3345 WinPython64-3.14.7.0dotf.exe 18 112 856 Bytes 19cb4a9830996c2801306061887c59a5e2638afd88979b43857011b8e2fd7662
    51b2fba7e3da5d67777203ebd93d3fd3 e2647ce7c7368697a451451d020db98ddef4d7fd 41fe675059f5c9fa287b2c4e8820b5f012227c7b91e574b0fa082f42abb2c45a WinPython64-3.14.7.0dotf.zip 28 889 568 Bytes bfedc707caf6ddb9dd724c1eaa697891fce5a2a69a3f1eca5775235484c1ed2f
    b49db5a64e73626cd56259e0c5af99f6 ae49fdd0411622627168704d41916d8356ba3c27 e0cb2bbc4c261cce9be1cf47482200325cb6a95aaa4340fc135243434f81e9ca WinPython64-3.14.7.0slim.7z 671 442 247 Bytes dbb79eea2fa52b4eeda131a0fc484acded5108e1903cfe2210fc5f9aec9e0449
    17e3f6d2fac1c34633aa7957c24c417e 67c8f946a8f56eb78ab030747dc20cb5858ac57e 22da1272004ca40c5471b21e8af9baa27db6531b2f0fdd64e0d15e0a0ac61e2f WinPython64-3.14.7.0slim.exe 671 657 643 Bytes 9161fa6a4ecb223ccdc6ae2dc512ed63d71269fb693197bc15dfdabf52b74601
    54a0cf34b83509ee1a3a26a8ef4aa9ad 8b5f3f80ca726cde3460b2f56b97a44bf9edccc2 bdefb0b935139649fa10fb6b2495dd9ecf33892684e823d0498175f3812a05d0 WinPython64-3.14.7.0slimf.7z 430 710 863 Bytes 1da043e9c46570ec67bf2c666dc3534ee8c170841b4f7f01c332052a136b20a9
    6829fbe9b470ca89b7798583ce939da0 9b548ca1665f34a23e9ffdda0b15d4976a186799 093f2cdafdc69484d720d350ffc33032109149a79343f027a21e4df6f887204b WinPython64-3.14.7.0slimf.exe 430 925 783 Bytes 00e3bc9a7db98e08c3071341fddfcd2378fde898959a04c630a9c1046e9e913c
    c3f0259f10ccbf09aaf711edf124a582 9a199ccfdf4981e948bae935354741c0aff4ccfd 42bfa45545a2ca5384da630f11046176903c483b2a918af9b3d23f6d29b918b2 WinPython64-3.15.0.4dot.exe 18 843 092 Bytes f65a07febb56a2a1b96128a0785fb92253eb93a7b55e985c66b0eb0bf2e82d96
    f86a63c0cd18d99bb1702a90f137661b b1c98d17b0df105338dfb87edb63b1337ab17617 cf3619c8bf1f1db91931359309c3c8cd62f2b1917b935b910379bf5c7948d56b WinPython64-3.15.0.4dot.zip 29 872 618 Bytes ca0b1b778cc32678a4e0cad3c8abfc627eba4c2c1c18a15ad9bf116f44ff2a78

    pylock and requirements versions: (now also in the winpython/changelogs comits), and how-to use with pip:
    . with a requirement file:
    . pip download --dest C:\your_temp_Wheelhouse --no-deps --require-hashes -r requir.64-3_14_7_0slim.txt
    . pip install --no-deps --no-index --trusted-host=None --find-links=C:\your_temp_Wheelhouse --require-hashes -r requir.64-3_14_7_0slim.txt
    . with a pylock file:
    . pip install --no-deps --require-hashes -r pylock.64-3_14_7_0slim.toml --no-warn-script-location

    MD5 SHA-1 SHA-256 Binary Size blake2b-256
    3ded77f22dfbc7c3c4087890c52944fd 18a6477deabe194fd05f19c6452aa3c0a52bee8f 7e01f782094ee309f2b319036bd39e7a914ae0407b50daf8f4be5a466bdb78d0 pylock.64-3_13_15_0dot.toml 3 062 Bytes 47912cf7cd4f5b5a2c9062b92f58353bec91d7eda5aae35a699e7e60b1a7e044
    8bd2a8b3b45254dcf76bbf72d93f26b0 58c54303aab1201385f42f14c86cb28690873033 dfe5355c8c35c5e2f7736a3965c039ff3d613ca7340551e0ddd13f5bdcbf007b pylock.64-3_13_15_0slim.toml 203 475 Bytes 097b6b5c4c51888ed23f243785aeef222b063d84a627bc15e2756d061ea76c20
    3ded77f22dfbc7c3c4087890c52944fd 18a6477deabe194fd05f19c6452aa3c0a52bee8f 7e01f782094ee309f2b319036bd39e7a914ae0407b50daf8f4be5a466bdb78d0 pylock.64-3_14_7_0dot.toml 3 062 Bytes 47912cf7cd4f5b5a2c9062b92f58353bec91d7eda5aae35a699e7e60b1a7e044
    3ded77f22dfbc7c3c4087890c52944fd 18a6477deabe194fd05f19c6452aa3c0a52bee8f 7e01f782094ee309f2b319036bd39e7a914ae0407b50daf8f4be5a466bdb78d0 pylock.64-3_14_7_0dotf.toml 3 062 Bytes 47912cf7cd4f5b5a2c9062b92f58353bec91d7eda5aae35a699e7e60b1a7e044
    65170025cc54cc2596018cd9631d7d6d 312e23a6d3df2a9b2d26817d1bf4e54d481ff0a2 4c51739e0b05dcdab614f08b3b8a2fe8fa39b7db00f040548fff8e2cba1594aa pylock.64-3_14_7_0slim.toml 203 457 Bytes b0246ffd254fad44b2775368468997cbb20cbfaf6049b23367980a0dc88e9b99
    234a02778b999a92485c8d8b8b07a792 b2311ca3bb61af6fc430db98831edd3fea7898cd db61fb21f8189055ce9c64891c9a4873f8e4c85a2624c28d363071bd8319a319 pylock.64-3_14_7_0slimf.toml 180 290 Bytes 9e95776119c580804558ea08ea929e41cf9f815830f540e530376aeaa98274c0
    3ded77f22dfbc7c3c4087890c52944fd 18a6477deabe194fd05f19c6452aa3c0a52bee8f 7e01f782094ee309f2b319036bd39e7a914ae0407b50daf8f4be5a466bdb78d0 pylock.64-3_15_0_4dot.toml 3 062 Bytes 47912cf7cd4f5b5a2c9062b92f58353bec91d7eda5aae35a699e7e60b1a7e044
    875310d2ced61c055187299259f33cbe 8c43bc51807a7ac587cdc15c9e52fded7c68ab57 85a1093793204d49296339650bf23d51e56fd3a1accd931a58ad4592d15d44d1 requir.64-3_13_15_0dot.txt 836 Bytes a0d6fd0f56d5f5a2bb78d25ffe5258777a175daa4f3ee35e0518a3d95ba5a61c
    a790ee1fee2cf5c5aba6f7b46f411d96 e5f45cc192d27a5b1efa40ecd799195cfd6c4f29 4d5e6b1126afaf1cc52e12be59021b08be2917e5590ff6aec6ef337657717e25 requir.64-3_13_15_0slim.txt 55 873 Bytes 12528174d33352946ea9b2b081b6fa1c5e0dcf2f9f1147381f954df12672df0a
    875310d2ced61c055187299259f33cbe 8c43bc51807a7ac587cdc15c9e52fded7c68ab57 85a1093793204d49296339650bf23d51e56fd3a1accd931a58ad4592d15d44d1 requir.64-3_14_7_0dot.txt 836 Bytes a0d6fd0f56d5f5a2bb78d25ffe5258777a175daa4f3ee35e0518a3d95ba5a61c
    875310d2ced61c055187299259f33cbe 8c43bc51807a7ac587cdc15c9e52fded7c68ab57 85a1093793204d49296339650bf23d51e56fd3a1accd931a58ad4592d15d44d1 requir.64-3_14_7_0dotf.txt 836 Bytes a0d6fd0f56d5f5a2bb78d25ffe5258777a175daa4f3ee35e0518a3d95ba5a61c
    b7f9934ff638f50bb79612669ee17ec8 860fac79e4f9a48d52d6f982fe90ea5378ef515f d8e4bd25ebe3bc89d73ed174e368810b893273acccff6ea0d1c71bca1f461279 requir.64-3_14_7_0slim.txt 55 873 Bytes 312f421f5188adb95d02ffea94d9b2464c77ed64c7e3f6addf312dbf480193f0
    0114038d6d020a948c86875ec09d9a05 e3c5aeda9d59eb5e3804351c1e1ec5421b22b1e6 81ea7e251955eedb6222c3564286c7a5868b7387d9efd43bc2cee2b7defa56f8 requir.64-3_14_7_0slimf.txt 49 560 Bytes 781a2c0424a80ff2e9d18806a4c85fb4b96e87d3c10a45468be8ce808580825d
    875310d2ced61c055187299259f33cbe 8c43bc51807a7ac587cdc15c9e52fded7c68ab57 85a1093793204d49296339650bf23d51e56fd3a1accd931a58ad4592d15d44d1 requir.64-3_15_0_4dot.txt 836 Bytes a0d6fd0f56d5f5a2bb78d25ffe5258777a175daa4f3ee35e0518a3d95ba5a61c
  9. stonebig
    added this to the 2026-03 towards AI local / 3.15.0rc / jupyterlite milestone on Aug 23, 2026

  10. stonebig commented on Aug 27, 2026

    @stonebig
    stonebig
    on Aug 27, 2026
    ContributorAuthor

    still need to update the website , using claude. by the way , DDH (of Ruby-on-Rails) has spoken about AI: https://youtu.be/NYFGCESmikA

    DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux | Lex Fridman Podcast

    ... and you should listen (5 hours)

    previous interview, he was against, but since this winter, things have changed totally .... but listen by yourself

  11. stonebig commented on Aug 28, 2026

    @stonebig
    stonebig
    on Aug 28, 2026
    ContributorAuthor

    So the website https://winpython.github.io/ is being redone this august 28th per my new Marketing Assistant, Claude, that I suppose:

    • is expert at it, like most AI,
    • but with some taste, for example no kitten image nor video of a disaster.

  12. stonebig commented on Aug 28, 2026

    @stonebig
    stonebig
    on Aug 28, 2026
    ContributorAuthor

    I'm pretty much in DHH feelings, we enjoyed the moment we switched from assembly to C.

  13. stonebig
    closed this as completedon Aug 28, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions