<?xml version='1.0' encoding='utf-8'?>
<rss version="2.0">
  <channel>
    <title>PlaintextML developer changes</title>
    <link>https://www.plaintextml.com/changes.html</link>
    <description>Selected developer changes and record revisions, backed by official documentation.</description>
    <item>
      <title>PyTorch 2.6 makes weights-only checkpoint loading the default</title>
      <link>https://www.plaintextml.com/changes/pytorch-2-6-torch-load-weights-only-default.html</link>
      <description>PyTorch 2.6 changed `torch.load` to behave as `weights_only=True` when callers do not provide a custom `pickle_module`. Plain tensor state dictionaries should continue loading, while serialized modules, tensor subclasses, NumPy objects, and other custom globals can now fail until the caller chooses an explicit trusted-load path or allowlists required types. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>breaking-change</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/pytorch-2-6-torch-load-weights-only-default.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>OpenAI Assistants API shut down in favor of Responses and Conversations</title>
      <link>https://www.plaintextml.com/changes/openai-assistants-api-sunset-2026.html</link>
      <description>OpenAI has shut down the Assistants API and directs integrations to the Responses API, with Conversations used for persistent state. The migration changes the core object model from Assistants, Threads, Runs, and Run Steps to Prompts, Conversations, Responses, and Items. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>deprecation</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/openai-assistants-api-sunset-2026.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>NumPy 2.0 breaks the 1.x binary ABI and removes legacy APIs</title>
      <link>https://www.plaintextml.com/changes/numpy-2-0-abi-and-api-break.html</link>
      <description>NumPy 2.0 introduced the project’s first major ABI break since the 1.x line, so extension binaries built against NumPy 1.x may fail to import under 2.0. It also removed old Python and C APIs and adopted new type-promotion behavior, creating migration work even for some projects without compiled code. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>breaking-change</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/numpy-2-0-abi-and-api-break.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>MLflow 3 changes model storage and removes deprecated APIs</title>
      <link>https://www.plaintextml.com/changes/mlflow-3-model-and-api-migration.html</link>
      <description>MLflow 3 promotes logged models to first-class entities, changes model logging from `artifact_path` to `name`, and stores model artifacts outside the run-artifact hierarchy. It also removes Recipes, several model flavors and parameters, the deployment-server command, and portions of the older evaluation API. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>breaking-change</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/mlflow-3-model-and-api-migration.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Transformers v5 removes TensorFlow and JAX backends and refactors core APIs</title>
      <link>https://www.plaintextml.com/changes/hugging-face-transformers-v5-migration.html</link>
      <description>Transformers v5 made PyTorch the sole model backend, removed long-deprecated surfaces, and replaced the slow-versus-fast tokenizer split with a new backend architecture. The major release also added dynamic weight conversion and moved to `huggingface_hub&gt;=1.0`, bringing HTTPX exceptions, environment-based proxy configuration, and cache-variable changes into the migration. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>breaking-change</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/hugging-face-transformers-v5-migration.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Gemini Interactions API replaced outputs with a typed steps timeline</title>
      <link>https://www.plaintextml.com/changes/google-gemini-interactions-steps-schema-2026.html</link>
      <description>Google replaced the Interactions API outputs array with typed steps and consolidated output controls under a polymorphic response_format. Its current migration timeline marks the legacy schema removed on June 8, 2026, after the new schema became the default on May 26. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>breaking-change</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/google-gemini-interactions-steps-schema-2026.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Gemini 2.0 Flash and Flash-Lite API endpoints were shut down</title>
      <link>https://www.plaintextml.com/changes/google-gemini-2-0-flash-shutdown-2026.html</link>
      <description>Google shut down the Gemini 2.0 Flash and Flash-Lite model IDs, including their -001 variants, on June 1, 2026. The current lifecycle table recommends Gemini 3.6 Flash for the Flash IDs and Gemini 3.1 Flash-Lite for the Lite IDs. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>deprecation</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/google-gemini-2-0-flash-shutdown-2026.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
    <item>
      <title>Newer Claude models restrict or reject explicit sampling parameters</title>
      <link>https://www.plaintextml.com/changes/anthropic-post-opus-4-6-sampling-parameters.html</link>
      <description>For models released after Claude Opus 4.6, Anthropic accepts `temperature` only at 1.0 and `top_p` only at 0.99 or higher, while any explicit `top_k` is rejected with HTTP 400. The Python SDK version 1.0 and later rejects these fields locally with TypeError. Source checked: 2026-09-07. Latest revision: Initial record added after checking the linked official documentation; migration checks are editorial interpretation.</description>
      <category>breaking-change</category>
      <guid isPermaLink="true">https://www.plaintextml.com/changes/anthropic-post-opus-4-6-sampling-parameters.html</guid>
      <pubDate>Mon, 07 Sep 2026 00:00:00 GMT</pubDate>
    </item>
  </channel>
</rss>