<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>RelationalAI Release Notes</title>
    <link>https://private.relational.ai/release-notes</link>
    <description>Latest updates for RelationalAI Release Notes.</description>
    <atom:link href="https://private.relational.ai/release-notes/rss/all.xml" rel="self" type="application/rss+xml" />
    <lastBuildDate>Thu, 19 Nov 2026 05:18:08 GMT</lastBuildDate>
    <item>
      <title>2.2.1</title>
      <link>https://private.relational.ai/release-notes/python/2.2.1</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/python/2.2.1</guid>
      <pubDate>Thu, 19 Nov 2026 05:18:08 GMT</pubDate>
      <category>Python package and CLI</category>
      <description>Optimization text no longer appears in solve_info.</description>
      <content:encoded><![CDATA[<p>Version 2.2.1 of the relationalai Python package is now available!</p>
<p>To upgrade, activate your virtual environment and run the following command:</p>
<p>pip install --upgrade relationalai</p>
<h2>New Features and Enhancements</h2>
<ul>
<li><a href="/build/reasoning/prescriptive/">Prescriptive reasoning</a> now has <a href="/api/python/v2.2/semantics/reasoners/prescriptive/problem/problem_class/index.md#method-install_display"><code>Problem.install_display()</code></a> to define the tables used to display a problem&#39;s decision variables, objectives, and constraints before deployment. Prescriptive reasoning is in Public Preview.</li>
</ul>
<h2>Bug Fixes</h2>
<ul>
<li><p>Fixed an issue that caused slow Snowflake refreshes for <a href="/deploy/deploy-a-model">deployed models</a> whose rules produce multiple properties and filter rows using a negated condition (<a href="/api/python/v2.2/semantics/not_/index.md"><code>not_()</code></a>) or a join used only for filtering.</p>
</li>
<li><p>Fixed an issue that caused <a href="/api/cli/v2.2/models/deploy/index.md"><code>rai models deploy</code></a> in <a href="/deploy/deploy-a-model">model deployment</a> to change Snowflake objects before reporting that a Snowflake schema created as a branch in the retired <a href="/release-notes/python/2.2.0/#deprecations-and-removals">Model Branching workflow</a> could not be deployed. It now stops with an unsupported-branch error and makes no changes to Snowflake objects, while still allowing upgrades of older models that were not created as branches in that workflow.</p>
</li>
<li><p>Fixed an issue that caused confusing GNN setup errors in <a href="/build/reasoning/predictive/">predictive reasoning</a> when training or validation data included a time column, or test data fields did not match. Errors now identify the <code>has_time_column=True</code> setting and list the expected test fields. For unlabeled rows, use <a href="/api/python/v2.2/semantics/reasoners/predictive/estimators/relational/gnn/gnn_class/index.md#method-predictions"><code>GNN.predictions()</code></a> instead of <code>test=</code>. Predictive reasoning is in Public Preview.</p>
</li>
<li><p>Fixed an issue that caused a <code>RelationshipNameMismatch</code> warning in <a href="/build/reasoning/predictive/">predictive reasoning</a> when the training and validation datasets had different names but the same fields.</p>
</li>
<li><p>Fixed an issue that caused incremental deployments in <a href="/deploy/deploy-a-model">model deployment</a> to recreate unchanged scheduled refresh tasks and lose their history.</p>
</li>
<li><p>Fixed an issue that caused integer exponentiation, such as <code>3 ** 4</code>, to fail in PyRel queries and definitions with the default settings.</p>
</li>
</ul>
<h2>Deprecations and Removals</h2>
<ul>
<li><a href="/build/reasoning/prescriptive/">Prescriptive reasoning</a> no longer includes a text version of your optimization problem in <a href="/api/python/v2.2/semantics/reasoners/prescriptive/problem/problem_class/index.md#method-solve_info"><code>Problem.solve_info</code></a> by default. The <a href="/api/python/v2.1/semantics/reasoners/prescriptive/problem/problem_class/index.md#method-printed_model"><code>Problem.printed_model</code></a> method and the <code>SolveInfoData.printed_model</code> field are removed. To retrieve or save the text, pass <code>print_format</code> to <a href="/api/python/v2.2/semantics/reasoners/prescriptive/problem/problem_class/index.md#method-solve"><code>Problem.solve</code></a> and use <a href="/api/python/v2.2/semantics/reasoners/prescriptive/problem/problem_class/index.md#method-read_printed_model"><code>Problem.read_printed_model</code></a> or <a href="/api/python/v2.2/semantics/reasoners/prescriptive/problem/problem_class/index.md#method-write_printed_model"><code>Problem.write_printed_model</code></a>. Print requests require Native App schema version 3, and older builds reject them and can fail deployed refreshes. The removed APIs have no replacement in model definitions or constraints.</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2026.11.14-9faf9f7-1</title>
      <link>https://private.relational.ai/release-notes/native-app/2026.11.14-9faf9f7-1</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/native-app/2026.11.14-9faf9f7-1</guid>
      <pubDate>Wed, 18 Nov 2026 00:00:00 GMT</pubDate>
      <category>Native App</category>
      <description>Query reasoner run times and cost estimates in SQL.</description>
      <content:encoded><![CDATA[<p>Version 2026.11.14-9faf9f7-1 of the RelationalAI Native App is now available!</p>
<p>Note that RelationalAI Native App upgrades are applied automatically and require no action on your part, unless you have opted-in to manual upgrades.</p>
<h2>New Features and Enhancements</h2>
<ul>
<li>Reasoner run durations and usage estimates are now available through three preview SQL functions: <a href="/api/sql/observability_preview/reasoner_run_duration"><code>REASONER_RUN_DURATION()</code></a>, <a href="/api/sql/observability_preview/reasoner_usage"><code>REASONER_USAGE()</code></a>, and <a href="/api/sql/observability_preview/reasoner_cost"><code>REASONER_COST()</code></a>. The estimates cover Snowflake credits, RelationalAI units, and costs in U.S. dollars. They are not billing records. To query them, <a href="/get-started/operate#enable-app-observability">enable app observability</a>, use an active Snowflake warehouse, and use the <code>observability_viewer</code> <a href="/api/sql/application-roles#service-roles">application role</a>.</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2.2.0</title>
      <link>https://private.relational.ai/release-notes/python/2.2.0</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/python/2.2.0</guid>
      <pubDate>Tue, 17 Nov 2026 23:30:09 GMT</pubDate>
      <category>Python package and CLI</category>
      <description>Queries to unpublished outputs fail in PyRel 2.2.0.</description>
      <content:encoded><![CDATA[<p>Version 2.2.0 of the relationalai Python package is now available!</p>
<p>To upgrade, activate your virtual environment and run the following command:</p>
<p>pip install --upgrade relationalai</p>
<h2>Breaking Changes</h2>
<ul>
<li><p><a href="/deploy/deploy-a-model">Model Deployment</a> now raises an <code>UnpublishedQueryOutput</code> error when a query reads a result that your model doesn&#39;t publish. For example, say you call <a href="/api/python/v2.2/semantics/reasoners/graph/core/graph_class/index.md#method-degree"><code>Graph.degree()</code></a> but never attach the result to a concept or use it in a definition. PyRel saves that result only in the model&#39;s internal schema. Earlier versions let queries read this internal copy. To fix the error, attach the result to a concept. Then deploy the model again with <a href="/api/cli/v2.2/models/deploy/index.md"><code>rai models deploy</code></a> and query the new property:</p>
<pre><code class="language-python"># Works: attach the result, deploy, and query it.
Account.degree = graph.degree()
m.select(Account.id, Account.degree).to_df()

# Raises UnpublishedQueryOutput: the result isn&#39;t attached.
degree = graph.degree()
count = Integer.ref()
m.where(degree(Account, count)).select(Account.id, count).to_df()
</code></pre>
<p>If you can&#39;t attach a result to a concept, use it in a definition instead. See <a href="/deploy/configure-deployment#expose-algorithm-results-as-outputs">Expose algorithm results as outputs</a>.</p>
</li>
</ul>
<h2>New Features and Enhancements</h2>
<ul>
<li><p><a href="/deploy/reasoners#manage-prescriptive-reasoners">Prescriptive Reasoning</a> and <a href="/deploy/reasoners#manage-predictive-reasoners">Predictive Reasoning</a> can now use the <a href="/api/python/v2.2/config/config_reasoners_fields/enginesize/index.md"><code>MEM_X64_G2_32</code></a> reasoner size on AWS and Azure. To use it, set <a href="/api/python/v2.2/config/config_reasoners_fields/reasonerconfig_class/index.md#param-size"><code>ReasonerConfig.size</code></a>. To see which sizes you can use, run <a href="/api/cli/v2.2/reasoners/sizes/index.md"><code>rai reasoners sizes</code></a>. Both features are in Public Preview. Prescriptive Reasoning is available by request.</p>
</li>
<li><p>New projects that you create with <a href="/api/cli/v2.2/models/init/index.md"><code>rai models init</code></a> no longer include the <code>auto_deploy</code> setting. Projects that already have the setting still work. If you leave it out, it defaults to <code>False</code>. If you query a model that isn&#39;t deployed, the error now tells you to run <a href="/api/cli/v2.2/models/deploy/index.md"><code>rai models deploy</code></a>. See <a href="/deploy/configure-deployment">Configure deployment</a>.</p>
</li>
<li><p>Models with many concepts now deploy faster. In one DuckDB test, a model with 480 entities deployed in 16 seconds instead of 31 seconds. Your results may differ.</p>
</li>
<li><p>With the SQL Reasoner, queries that join a table to itself and filter by a list now run faster. In one test with 4,000 items, a query took 0.46 seconds instead of 5.04 seconds. This was a single test, and the PyRel query was still slower than the same query in plain SQL.</p>
</li>
</ul>
<h2>Bug Fixes</h2>
<ul>
<li><p>Fixed an issue that caused <a href="/api/cli/v2.2/models/list/index.md"><code>rai models list</code></a> to fail on large Snowflake accounts. The command now checks up to 10,000 matches across your account. It warns you when the list might be incomplete. If you see this warning, a model that isn&#39;t listed might still exist. The <code>--limit</code> option only controls how many results you see. It doesn&#39;t raise the 10,000-match cap. Some older models that use only tags as their ID might not appear in the list.</p>
</li>
<li><p>Fixed an issue that caused <a href="/api/python/v2.2/semantics/std/aggregates/median/index.md"><code>aggregates.median()</code></a> and <a href="/api/python/v2.2/semantics/std/aggregates/percentile/index.md"><code>aggregates.percentile()</code></a> to fail when you passed a key before the value. These calls now use one value for each key, so repeated values count. For example, say four employees earn 10, 10, 10, and 40. Then <code>aggregates.median(Employee, Employee.salary)</code> returns 10. Without the key, <code>aggregates.median(Employee.salary)</code> counts each distinct value once and returns 25. Keyed calls also work with <a href="/api/python/v2.2/semantics/frontend/base/aggregate_class/index.md#method-per"><code>Aggregate.per()</code></a>, as in <code>aggregates.percentile(75, Employee, Employee.salary).per(Department)</code>.</p>
</li>
<li><p>Fixed an issue that caused <a href="/api/python/v2.2/semantics/std/aggregates/top/index.md"><code>aggregates.top()</code></a>, <a href="/api/python/v2.2/semantics/std/aggregates/bottom/index.md"><code>aggregates.bottom()</code></a>, <a href="/api/python/v2.2/semantics/std/aggregates/limit/index.md"><code>aggregates.limit()</code></a>, and <a href="/api/python/v2.2/semantics/std/aggregates/rank/index.md"><code>aggregates.rank()</code></a> to return different rows when values were tied. This happened with the SQL Reasoner. These functions now break ties with the values that identify each entity, so the same query returns the same rows. Two cases can still change the results. Floating-point sums that differ only by rounding can change the order. Recursive <code>top</code> queries can return more rows than the limit.</p>
</li>
<li><p>Fixed an issue that caused an <code>Unknown column</code> error when deploying with the SQL Reasoner. This happened when a concept was identified by more than one field. A definition that read a one-field relationship of that concept, such as <code>{Order} is flagged</code>, failed. This bug started in PyRel 2.0.0.</p>
</li>
<li><p>Fixed an issue that caused a count to return 0 for every entity when a definition compared it with the entity&#39;s own ID. This bug affected the SQL Reasoner in PyRel 2.1.1.</p>
</li>
<li><p>Fixed an issue that caused deployment to Snowflake to fail for a definition that filtered grouped aggregates with a union. A union matches rows that meet any of several conditions.</p>
</li>
<li><p>Fixed an issue that caused the order of selected columns to change which rows a query returned. This happened with the SQL Reasoner.</p>
</li>
<li><p>Fixed an issue that caused a query to return a related value when the relationship needed to reach it didn&#39;t exist. This happened with the SQL Reasoner.</p>
</li>
<li><p>Fixed an issue that caused a query to drop an entity that it matched through more than one relationship. This happened when the selected property was missing on one of those relationships. This bug affected the SQL Reasoner.</p>
</li>
<li><p>Fixed an issue that caused a query that selects only aggregates to return different results based on column order. Now, if any aggregate has no input values, the query returns no rows. To get a value instead, add a default, such as <a href="/api/python/v2.2/semantics/std/aggregates/sum/index.md"><code>aggregates.sum(Employee.salary)</code></a> <code>| 0.0</code>. This bug affected the SQL Reasoner.</p>
</li>
<li><p>Fixed an issue that caused definitions to treat <code>NULL</code> values in source tables as real values. The SQL Reasoner now treats them as missing.</p>
</li>
<li><p>Fixed an issue that caused Model Deployment with the Logic Reasoner to fail for some subtypes. This happened when a subtype inherited an ID made of more than one field. This fix addresses a <code>KeyError</code> reported in PyRel 2.1.1.</p>
</li>
<li><p>Fixed an issue that caused unclear errors when PyRel couldn&#39;t read a source table or column. The error now explains why. For a missing column, it names the table and lists the columns that PyRel knows about. If that list might be out of date, the error tells you to clear the schema cache. See <a href="/build/modeling/declare-data-sources">Declare data sources</a>.</p>
</li>
</ul>
<h2>Deprecations and Removals</h2>
<ul>
<li><p>Model Branching is retired in PyRel 2.2.0. Its legacy branches are Snowflake schemas, not source control branches. PyRel now rejects them when you pull or deploy. It doesn&#39;t move them for you. To keep using a model from a legacy branch, <a href="/deploy/deploy-a-model">deploy the full model to a new standalone schema</a> with <a href="/api/cli/v2.2/models/deploy/index.md"><code>rai models deploy</code></a>. Then run <a href="/api/cli/v2.2/models/teardown/index.md"><code>rai models teardown</code></a> to remove the legacy branch.</p>
</li>
<li><p>Your local project might have unfinished work from Model Branching. In that case, PyRel 2.2.0 won&#39;t overwrite your local state file. Instead, <a href="/api/cli/v2.2/models/pull/index.md"><code>rai models pull --force</code></a> and <a href="/api/cli/v2.2/models/deploy/index.md"><code>rai models deploy --force</code></a> fail. Don&#39;t edit or delete that file. <code>rai doctor</code> reports the state as unsupported. It recommends that you <a href="/deploy/deploy-a-model">deploy the full model to a new standalone schema</a>.</p>
</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2.1.1</title>
      <link>https://private.relational.ai/release-notes/python/2.1.1</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/python/2.1.1</guid>
      <pubDate>Fri, 13 Nov 2026 17:04:00 GMT</pubDate>
      <category>Python package and CLI</category>
      <description>New sizes for Logic and Prescriptive Reasoners.</description>
      <content:encoded><![CDATA[<p>Version 2.1.1 of the relationalai Python package is now available!</p>
<p>To upgrade, activate your virtual environment and run the following command:</p>
<p>pip install --upgrade relationalai</p>
<h2>New Features and Enhancements</h2>
<ul>
<li><p><a href="/deploy/reasoners#what-logic-reasoners-do">Logic Reasoners</a> now support <a href="/api/python/v2.1/config/config_reasoners_fields/enginesize"><code>MEM_X64_G2_32</code></a>, which you can select by setting <a href="/api/python/v2.1/config/config_reasoners_fields/reasonerconfig_class/index.md#param-size"><code>reasoners.logic.size</code></a> as shown in <a href="/deploy/configure-deployment#select-which-reasoners-to-use">Select which reasoners to use</a>. Availability depends on your account, cloud, and Snowflake instance family. Predictive and Prescriptive Reasoners do not support this size.</p>
</li>
<li><p><a href="/deploy/reasoners#manage-prescriptive-reasoners">Prescriptive Reasoners</a> now support <a href="/api/python/v2.1/config/config_reasoners_fields/awsenginesize"><code>GPU_NV_S</code></a> on AWS and <a href="/api/python/v2.1/config/config_reasoners_fields/azureenginesize"><code>GPU_NV_SM</code></a> on Azure. Set <a href="/deploy/configure-deployment#configure-prescriptive-reasoner-settings"><code>reasoners.prescriptive.size</code></a> to choose one. These reasoners are in Public Preview.</p>
</li>
</ul>
<h2>Bug Fixes</h2>
<ul>
<li><p>Fixed an issue that caused queries using <a href="/api/python/v2.1/semantics/frontend/base/model_class#method-union"><code>Model.union()</code></a> to omit matching rows.</p>
</li>
<li><p>Fixed an issue that caused <a href="/deploy/deploy-a-model#review-changes-before-deploying">model deployment</a> reports to classify routine telemetry updates incorrectly. Reports now list them as modifications.</p>
</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2.1.0</title>
      <link>https://private.relational.ai/release-notes/python/2.1.0</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/python/2.1.0</guid>
      <pubDate>Wed, 11 Nov 2026 01:45:12 GMT</pubDate>
      <category>Python package and CLI</category>
      <description>Retrain some GNN models after upgrading.</description>
      <content:encoded><![CDATA[<p>Version 2.1.0 of the relationalai Python package is now available!</p>
<p>To upgrade, activate your virtual environment and run the following command:</p>
<p>pip install --upgrade relationalai</p>
<h2>New Features and Enhancements</h2>
<ul>
<li><p><a href="/deploy/deploy-a-model">Model deployment</a> now shows what changed when a <a href="/deploy/deploy-a-model#review-changes-before-deploying">reviewed plan</a> is out of date. The &quot;Deployment refused&quot; error from <a href="/api/python/v2.1/semantics/frontend/base/model_class#method-deploy"><code>Model.deploy()</code></a> and <a href="/api/cli/v2.1/models/deploy"><code>rai models deploy --reviewed-plan</code></a> now includes a line-by-line diff between your reviewed plan and the current plan, so you no longer have to compare plan files yourself.</p>
</li>
<li><p><a href="/build/reasoning/predictive">Predictive reasoning</a> now fails right away, instead of retrying for several minutes, when your <a href="/deploy/configure-deployment#set-the-predictive-reasoner-name-and-size"><code>reasoners.predictive</code> settings</a> in <code>raiconfig.yaml</code> don&#39;t match an existing predictive reasoner. Out-of-memory errors from training or generating predictions now also suggest a <a href="/deploy/configure-deployment#which-compute-pools-host-predictive-reasoners">larger predictive reasoner size</a>.</p>
</li>
</ul>
<h2>Bug Fixes</h2>
<ul>
<li><p>Fixed an issue that caused a registered Graph Neural Network (GNN) model in <a href="/build/reasoning/predictive">predictive reasoning</a> to sometimes stop working after you added another <a href="/api/python/v2.1/semantics/reasoners/graph/core/graph_class"><code>Graph</code></a> to the same PyRel model. After you upgrade, <a href="/build/reasoning/predictive/configure-a-GNN">retrain</a> and <a href="/build/reasoning/predictive/register-a-model">re-register</a> each GNN model registered with an earlier PyRel version, unless you passed your own edge concept to its <code>Graph</code> with <code>edge_concept=</code>.</p>
</li>
<li><p>Fixed an issue that caused top N and ranking queries that sort from highest to lowest, using <a href="/api/python/v2.1/semantics/std/aggregates/limit"><code>limit()</code></a> or <a href="/api/python/v2.1/semantics/std/aggregates/rank"><code>rank()</code></a> with <a href="/api/python/v2.1/semantics/std/aggregates/desc"><code>desc()</code></a>, to return missing (<code>NULL</code>) values first in their results when <a href="/deploy/deploy-a-model">model deployment</a> is enabled. The SQL that PyRel sends to Snowflake now sorts missing values last. If a deployed model&#39;s outputs use these queries, <a href="/deploy/deploy-a-model#deploy-from-the-cli">deploy the model again</a> with PyRel 2.1.0 to pick up the fix.</p>
</li>
<li><p>Fixed an issue that caused <a href="/deploy/collaborate-on-a-deployed-model#tear-down-a-model">tearing down a deployed model</a> with <a href="/api/cli/v2.1/models/teardown"><code>rai models teardown</code></a> or <a href="/api/python/v2.1/semantics/frontend/base/model_class#method-teardown"><code>Model.teardown()</code></a> to fail in some Snowflake accounts. Teardown failed with an &quot;Insufficient privileges&quot; error when the account had models deployed by roles whose tables your role can&#39;t read, and it also failed in databases with more than 10,000 schemas. Teardown now checks only the model you&#39;re tearing down, so it&#39;s also faster in large accounts.</p>
</li>
<li><p>Fixed an issue that caused interrupting a PyRel query or deployment to stop PyRel but not the SQL that PyRel had started in Snowflake. With <a href="/deploy/deploy-a-model">model deployment</a> enabled, this happened when you pressed Ctrl-C or interrupted a notebook cell. That SQL could keep running in Snowflake and using your warehouse. PyRel now also asks Snowflake to stop that SQL. Interrupt only once, because a second interrupt can leave the SQL running. To confirm that it stopped, check your Snowflake query history.</p>
</li>
<li><p>Fixed an issue that caused the <a href="/deploy/reasoners#what-reasoners-do">logic reasoner</a> to keep working on a deployed model&#39;s <a href="/deploy/configure-deployment#schedule-output-refreshes">output refresh</a> after the refresh was cancelled, including when you tore down or redeployed the model. The work kept running until it timed out. It now stops on its own, usually about a minute after the refresh is cancelled. To get this fix for a model deployed with an earlier PyRel version, <a href="/deploy/deploy-a-model#deploy-from-the-cli">deploy it again</a> with PyRel 2.1.0.</p>
</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2026.11.7-f8da33d</title>
      <link>https://private.relational.ai/release-notes/native-app/2026.11.7-f8da33d</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/native-app/2026.11.7-f8da33d</guid>
      <pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate>
      <category>Native App</category>
      <description>Minor improvements</description>
      <content:encoded><![CDATA[<p>Version 2026.11.7-f8da33d of the RelationalAI Native App is now available!</p>
<p>Note that RelationalAI Native App upgrades are applied automatically and require no action on your part, unless you have opted-in to manual upgrades.</p>
<h2>New Features and Enhancements</h2>
<ul>
<li>Minor improvements to performance and stability.</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2.0.1</title>
      <link>https://private.relational.ai/release-notes/python/2.0.1</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/python/2.0.1</guid>
      <pubDate>Fri, 06 Nov 2026 16:16:11 GMT</pubDate>
      <category>Python package and CLI</category>
      <description>Redeploy to restore deleted Snowflake objects.</description>
      <content:encoded><![CDATA[<p>Version 2.0.1 of the relationalai Python package is now available!</p>
<p>To upgrade, activate your virtual environment and run the following command:</p>
<p>pip install --upgrade relationalai</p>
<h2>New Features and Enhancements</h2>
<ul>
<li><p><a href="/deploy/deploy-a-model">Model Deployment</a> now supports full redeployment with <a href="/api/cli/v2.0/models/deploy"><code>rai models deploy --redeploy-all</code></a>. The command recreates all managed tables, views, and dynamic tables, even when their model definitions have not changed. This lets you restore objects deleted directly in Snowflake. Redeployment may rebuild materialized data and use Snowflake resources.</p>
</li>
<li><p>Model deployment schedules can now use a selected Snowflake warehouse for their refresh task and managed dynamic tables. Set <a href="/deploy/configure-deployment#schedule-output-refreshes"><code>deployment.schedules.&lt;name&gt;.refresh_warehouse</code></a> to choose an existing warehouse. If unset, the schedule uses the warehouse used to deploy the model. This setting does not create a warehouse.</p>
</li>
<li><p>Deployment reports now show separate counts for your model&#39;s outputs (tables, views, and dynamic tables) and the internal objects PyRel uses to manage deployment, plus the total. If your scripts read the JSON report, update them to use <code>summary.public</code>, <code>summary.internal</code>, and <code>summary.total</code> instead of the former combined count.</p>
</li>
<li><p>PyRel deployment plans now print as readable Markdown with <a href="/api/python/v2.0/semantics/frontend/base/model_class#method-plan_deployment"><code>Model.plan_deployment()</code></a>.</p>
</li>
</ul>
<h2>Bug Fixes</h2>
<ul>
<li><p>Fixed an issue that caused loading a PyRel model file that uses <a href="/api/python/v2.0/semantics/define"><code>define()</code></a> or <a href="/api/python/v2.0/semantics/select"><code>select()</code></a> to raise an “Ambiguous model” error when another model was already active in the Python session.</p>
</li>
<li><p>Fixed an issue that caused PyRel&#39;s <a href="/api/python/v2.0/semantics/std/aggregates/sum"><code>sum()</code></a>, <a href="/api/python/v2.0/semantics/std/aggregates/min"><code>min()</code></a>, and <a href="/api/python/v2.0/semantics/std/aggregates/max"><code>max()</code></a> queries to truncate results when the aggregate result type was wider than its inputs.</p>
</li>
<li><p>Fixed an issue that caused recursive PyRel rules using <code>min()</code> or <code>max()</code> to keep rows those aggregates had replaced.</p>
</li>
<li><p>Fixed an issue that caused a performance regression in a PyRel query using <a href="/api/python/v2.0/semantics/std/strings/split_part"><code>split_part()</code></a> at a fixed index inside an aggregate that joins a table to itself.</p>
</li>
<li><p>Fixed an issue that caused model deployment refreshes for deployed PyRel models with several metrics and separate filters to combine rows across metrics. Refreshes now keep each metric&#39;s rows separate.</p>
</li>
<li><p>Fixed an issue that caused model deployment to mishandle Snowflake schema and object names that require quoting, such as names with spaces or reserved words, during deployment, refresh, redeployment, permission checks, and teardown.</p>
</li>
<li><p>Fixed an issue that caused the PyRel debugger&#39;s refresh view to miss the refresh plan and log for configured model names with prefixes or suffixes, including when the Snowflake schema used to track the deployment has a name that must be quoted.</p>
</li>
<li><p>Fixed an issue that caused PyRel&#39;s status check for a refreshed Snowflake dynamic table to fail when its timestamp used a custom format with fractional seconds and a time-zone offset.</p>
</li>
<li><p>Fixed an issue that caused PyRel&#39;s default query mode to print an ignored cleanup exception at exit when model deployment was the first operation to open the Snowflake session and <a href="/api/python/v2.0/config/config_fields/modelconfig_class"><code>model.keep</code></a> was disabled.</p>
</li>
<li><p>Fixed an issue that caused PyRel to fail when saving <a href="/deploy/collaborate-on-a-deployed-model#how-deployments-are-tracked">model history</a> for a model with a property type that became more specific because of a uniqueness constraint or a calculation across a relationship path.</p>
</li>
<li><p>Fixed an issue that caused <a href="/build/reasoning/prescriptive">prescriptive reasoning</a> to accept decision-variable constraints using a list of allowed values, a union of values, or multiple alternatives with <code>|</code> under a different meaning. These forms now fail validation. Prescriptive reasoning is in Public Preview.</p>
</li>
<li><p>Fixed an issue that caused <a href="/deploy/deploy-a-model">model deployment</a> to show an identifier warning even when a deployment did not require a logic reasoner.</p>
</li>
<li><p>Fixed an issue that caused a PyRel query to fail when a mathematical expression received an input it could not use, such as a negative number for a square root or zero as a divisor. The affected row now has no value for that expression, and the query returns instead of failing.</p>
</li>
<li><p>Fixed an issue that caused some grouped <a href="/api/python/v2.0/semantics/std/aggregates/percentile"><code>percentile()</code></a> and <a href="/api/python/v2.0/semantics/std/aggregates/median"><code>median()</code></a> queries to fail.</p>
</li>
<li><p>Fixed an issue that caused <a href="/build/reasoning/prescriptive">prescriptive reasoning</a> constraints to skip entities or groups selected by <code>where()</code>, which could lead to incorrect solver results. Constraints now check every selected entity or group.</p>
</li>
<li><p>Fixed an issue that caused some values to be missing from multi-argument prescriptive reasoning decision variables.</p>
</li>
<li><p>Fixed an issue that caused <a href="/api/python/v2.0/semantics/reasoners/prescriptive/problem/problem_class#method-solve"><code>Problem.solve()</code></a> to silently drop an objective calculation with no value and allowed a second objective to fail later or be ignored. It now raises an error for an objective with no value and rejects a second objective when it is added.</p>
</li>
<li><p>Fixed an issue that caused PyRel queries to repeat the same join when selecting several properties from one related entity.</p>
</li>
<li><p>Fixed an issue that caused a property from a newly created related entity in a PyRel rule to be written as <code>NULL</code>.</p>
</li>
<li><p>Fixed an issue that caused Snowflake queries to fail or omit rows when using <a href="/api/python/v2.0/semantics/std/strings/split_part"><code>split_part()</code></a> with a literal or already-bound index in a negated filter or <code>| default</code> expression.</p>
</li>
<li><p>Fixed an issue that caused the PyRel CLI to show a generic error when a reviewed deployment plan was out of date. It now displays a formatted diagnostic and a clearer recovery message.</p>
</li>
<li><p>Fixed an issue that caused PyRel to skip whole-number values when loading integer data columns in source tables with missing cells. It now preserves those values and loads missing cells as null.</p>
</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2026.10.30-d68bfbc</title>
      <link>https://private.relational.ai/release-notes/native-app/2026.10.30-d68bfbc</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/native-app/2026.10.30-d68bfbc</guid>
      <pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate>
      <category>Native App</category>
      <description>Internal indexes for deployed models are now automatically cleaned up.</description>
      <content:encoded><![CDATA[<p>Version 2026.10.30-d68bfbc of the RelationalAI Native App is now available!</p>
<p>Note that RelationalAI Native App upgrades are applied automatically and require no action on your part, unless you have opted-in to manual upgrades.</p>
<h2>New Features and Enhancements</h2>
<ul>
<li>After this update, the internal indexes used to build output objects in a <a href="/build/guides/modeling/deploy-a-model">deployed model</a> will be rebuilt and scheduled for cleanup if they are not accessed again, such as during refresh or a re-deployment. If cleanup removes the data, a later refresh will rebuild these indexes automatically, and remove them again after nine days of inactivity. Model deployment is in Public Preview.</li>
</ul>]]></content:encoded>
    </item>
    <item>
      <title>2.0.0</title>
      <link>https://private.relational.ai/release-notes/python/2.0.0</link>
      <guid isPermaLink="true">https://private.relational.ai/release-notes/python/2.0.0</guid>
      <pubDate>Tue, 03 Nov 2026 23:16:18 GMT</pubDate>
      <category>Python package and CLI</category>
      <description>Model deployment requires a breaking migration.</description>
      <content:encoded><![CDATA[<p>Version 2.0.0 of the relationalai Python package is now available!</p>
<p>To upgrade, activate your virtual environment and run the following command:</p>
<p>pip install --upgrade relationalai</p>
<h2>Breaking Changes</h2>
<ul>
<li>Before redeploying a PyRel 1.x model with PyRel 2.0, update its <a href="/api/python/v2.0/config/config_deployment_fields/deploymentconfig_class">deployment configuration</a> and update SQL queries that read the model to use published views instead of internal objects. PyRel 2.0 rejects the old settings. Some existing op logs that record deployment history are incompatible with PyRel 2.0. If deployment reports an incompatible op log, <a href="/deploy/collaborate-on-a-deployed-model#tear-down-a-model">follow the teardown and redeployment steps</a>. Model deployment is in Public Preview.</li>
</ul>
<h2>New Features and Enhancements</h2>
<ul>
<li><p>Model deployment history, called an <a href="/deploy/collaborate-on-a-deployed-model#how-deployments-are-tracked">op log</a>, now uses standard Snowflake tables by default, including on accounts without Hybrid Table support. To use Hybrid Tables for faster appends on supported accounts, set <a href="/api/python/v2.0/config/config_fields/oplogconfig_class"><code>oplog.table_type</code></a> to <code>hybrid</code> before deployment. Changing the table type for an existing deployment requires deleting its op log table and redeploying.</p>
</li>
<li><p>You can now call <a href="/api/python/v2.0/semantics/frontend/base/model_class#method-plan_deployment"><code>Model.plan_deployment()</code></a> to review the changes PyRel plans to make in Snowflake instead of using the CLI&#39;s <a href="/api/cli/v2.0/models/deploy"><code>--plan</code> option</a>. The method returns a report without installing or deploying the model.</p>
</li>
<li><p>After you review the changes PyRel plans to make, you can require the deployment to match them. Pass the report from <a href="/api/python/v2.0/semantics/frontend/base/model_class#method-plan_deployment"><code>Model.plan_deployment()</code></a> to <a href="/api/python/v2.0/semantics/frontend/base/model_class#method-deploy"><code>Model.deploy(reviewed_plan=...)</code></a>, or save it and pass its file with <code>--reviewed-plan</code> to <a href="/api/cli/v2.0/models/deploy"><code>rai models:deploy</code></a>. If the planned changes differ, PyRel stops before applying them, though setup may already have updated deployment metadata.</p>
</li>
<li><p><a href="/deploy/deploy-a-model">Model deployment</a> now treats the model defined in your Python code as the complete model by default. Missing definitions can delete existing Snowflake objects, and outdated code can undo newer changes. Use this default only for a main model managed by one pipeline and set <a href="/api/python/v2.0/config/config_fields/oplogconfig_class"><code>oplog.stateless_deploy</code></a> to <code>false</code> for branches or collaboration.</p>
</li>
<li><p>Model deployment now warns and continues after some errors caused by damaged or incompatible deployment history. A later deployment rebuilds that history from the current Python model, and automatic recovery is enabled by default. To prevent new recovery requests, set <a href="/api/python/v2.0/config/config_fields/oplogconfig_class"><code>oplog.auto_recover</code></a> to <code>false</code>. Recovery does not repair local edits, and installation errors or other hard errors still stop deployment.</p>
</li>
<li><p>Deployment plan reports now label SQL to show whether PyRel plans to add, replace, modify, or drop a Snowflake object, when that information is available.</p>
</li>
<li><p><a href="/build/reasoning/graph">Graph Reasoning</a> results can now appear as columns in Snowflake when attached to a top-level concept or used in a definition. Graph Reasoning is in <a href="/release-notes/preview-features">Public Preview</a>.</p>
</li>
</ul>
<h2>Bug Fixes</h2>
<ul>
<li><p>Fixed an issue that caused an unclear error when a query used <a href="/api/python/v2.0/semantics/frontend/base/model_class#method-where"><code>Model.where()</code></a> without specifying what <a href="/api/python/v2.0/semantics/frontend/base/model_class#method-select"><code>Model.select()</code></a> should return. The error now suggests returning a Boolean or selecting values for matching rows.</p>
</li>
<li><p>Fixed an issue that caused failed deployments to leave Snowflake queries running, risking overwrites of newer results. When PyRel can access Snowflake&#39;s record of queries run in the current session, it cancels any that are still running before the next deployment starts.</p>
</li>
<li><p>Fixed an issue that caused the last successfully deployed model to become unavailable for queries when a later deployment failed during preparation or finalization.</p>
</li>
<li><p>Fixed an issue that caused SQL queries checking dynamic table refresh status to fail or time out. This happened during deployment for Snowflake tables with quoted names containing apostrophes or backslashes.</p>
</li>
<li><p>Fixed an issue that caused CLI help summaries to stop at 45 characters even on wider terminals. <a href="/api/cli/v2.0/commands"><code>rai --help</code></a> and <a href="/api/cli/v2.0/models"><code>rai models --help</code></a> now use the available width, subject to the formatter&#39;s limit.</p>
</li>
<li><p>Fixed an issue that caused <a href="/api/cli/v2.0/models/list"><code>rai models:list</code></a> to fail when your Snowflake permissions let you see a model record in another database but not read it. The command now warns and skips that record.</p>
</li>
<li><p>Fixed an issue that caused Snowflake objects to return rounded values after a property&#39;s type changed. Deployment now applies the type change before rebuilding objects that read the property through a view, even when the view&#39;s SQL definition has not changed.</p>
</li>
<li><p>Fixed an issue that caused queries on models without deployment configuration to raise <code>AttributeError</code> when PyRel looked up the database name. They now use the existing model-name fallback for that lookup.</p>
</li>
<li><p>Fixed an issue that caused a deployment plan to report changes and suggest redeployment when the model&#39;s outputs and destinations had not changed.</p>
</li>
<li><p>Fixed an issue that caused the <a href="/api/cli/v2.0/debugger"><code>rai debugger</code></a> Refresh Log tab to show dark panels on a light page. It now follows the selected theme.</p>
</li>
<li><p>Fixed an issue that caused Snowflake queries joining a table created by model deployment to itself to return rows that should have been filtered out. Deployment now marks primary keys with <code>NORELY</code> so Snowflake keeps the filters instead of removing them based on the key. The keys remain declared, and existing tables receive this setting the next time you deploy the model.</p>
</li>
<li><p>Fixed an issue that caused model deployment to lowercase field names generated from entity names even when configured to keep the original capitalization of column names. Deployment now preserves capitalization, such as publishing <code>Strategy</code> as <code>&quot;Strategy&quot;</code>, and normalizes reserved names such as <code>group</code> to <code>GROUP</code>. Update SQL and policy references that use the old quoted lowercase names after redeployment.</p>
</li>
<li><p>Fixed an issue that caused PyRel queries using <a href="/api/python/v2.0/semantics/std/math/round"><code>math.round()</code></a> to fail type checking when a numeric expression had an explicit integer type.</p>
</li>
</ul>]]></content:encoded>
    </item>
  </channel>
</rss>