AI-native Integration News

Mainstream support for Talend 7.3 ends on November 30, 2024. Purchased extended support runs through December 2026 with no renewal option; it covers only Severity 1 issues and excludes fixes for unsupported third-party components. The only migration path is Talend 8 or Qlik Talend Cloud.

Current stable patches: 8.5.10 (27.08.2026) and 8.4.24 (30.07.2026). Both branches are in active support (8.4 until the end of 2026, 8.5 until the end of 2027).

Improvements in library dependency management, API specification versioning, and flows with scheduler triggers. The goal is to reduce friction in the integration development cycle.

Fine-tuning guide for MultiVectorEncoder (v6.0): unsupervised checkpoints adapt to a new domain better than off-the-shelf models, and CachedMultiVectorMultipleNegativesRankingLoss decouples batch size from GPU memory. Fine-tuning mLateOn-medical achieved 0.9139 NDCG@10 (+0.062 over the best general-purpose retriever), with training taking 14.5 hours on a single RTX 3090.

Earlier, running AI Assistant in a self-hosted setup required manually enabling the instance-ai module and configuring the sandbox, model, and web search through env variables. In 2.35, the module is enabled by default, setup is one command, and the model provider (Anthropic, OpenAI, OpenRouter, or any OpenAI-compatible endpoint, including a local one), sandbox, and web search (native provider support or fallback to Brave Search/self-hosted SearXNG) are configured directly in the UI.

Customers can now be segmented and carry structured data via the same attribute system already used for products and models; new MANAGE_CUSTOMER_TYPES_AND_ATTRIBUTES permission gates it.

The March roadmap has been expanded from 4 to 5 priorities. Agent identity and SDK developer experience have been added as a separate item; messaging primitives (Tasks, subscriptions/listen, progress) are being extended to server-initiated events via webhooks/channels instead of polling, and the local stdio transport is proposed to be unified with remote HTTP. The release follows the 2026-07-28 specification, the largest revision since the protocol launched.

A production RAG infrastructure example: PostgreSQL full-text plus semantic search with Qwen3-Embedding-0.6B (256-dimensional Matryoshka vectors), corpus embedding in batches on HF Jobs (L4, about 75 articles/sec), Storage Buckets as a versioned layer between compute and the database, and merging via Reciprocal Rank Fusion with a lexical-only fallback on timeout.

The new MultiVectorEncoder class brings ColBERT-like late interaction on top of dense/sparse models, already in the library. It is more precise on nuances (token-to-token MaxSim instead of compressing into a single vector), but the index is orders of magnitude heavier: 311.5 MB vs 7.5 MB dense on 4,874 Natural Questions passages. 40+ open checkpoints (LateOn, mLateOn, ColBERT, ColPali) are available on HF Hub.

Desktop app (macOS/Windows/Linux) with two modes: Cowork handles routine tasks (summaries, dashboards, team workload analysis, weekly reports in 8 minutes instead of 4 hours manually), while Code builds and deploys apps/bots/automations from text prompts. It runs on the BitrixGPT 5.6 Agent 1M model; the weights are closed, with access only via subscription/API, tagged as closed-weights. Data is processed under the user's permissions and stored in CIS infrastructure.

Released three weeks after 3.6 Flash, with the release pace accelerating to almost weekly. Gains on real-world software-engineering benchmarks, with a focus on coding and agent tasks. Weights are closed, accessible only through Google API/cloud.

A single visibility and control layer for agents in Anypoint Platform: verifiable agent identity, isolation of provider credentials, emergency agent shutdown, and auditing. Runs on top of AWS, Azure, and F5.

FinOps tools for AI: request cost calculation at the actual rate, budget limits per agent, and expense breakdown by agent. Model Proxy is in GA; the remaining features (Cost Optimization, Model Wallet, Unified Dashboard) will follow later in the quarter.

In TAdviser’s MDM market review (about 900 projects, market size of 6-7 billion rubles by the end of 2025), most MDM deployments in CIS were attributed to Datareon solutions. This is TAdviser’s estimate, not an independent market share audit.

Routine maintenance releases: 3.14.7 includes about 499 bug fixes from 86 contributors, and 3.13.15 includes about 400 bug fixes. A regular support cycle with no announced critical security fixes.

New opt-in columnar storage skips the inverted index for analytical/log workloads. VectorDB index mode auto-configures quantization, preconditioning and oversampling for vector search — no manual tuning. Security adds agent-assisted alert triage ("Alert Zero").

The first release candidate of 3.15 is out, with the final release planned for 2026-10-01. Confirmed features include lazy imports (PEP 810), built-in frozendict (PEP 814), a Sentinel type (PEP 661), unpacking in comprehensions (PEP 798), and a new sampling profiler (PEP 799); JIT has added about 8-9% performance on x86-64 Linux.

Django dependency upgraded to 5.2.17, fixing CVE-2026-15307 (server-side file-write/request forgery via spatial lookups), CVE-2026-15337 and CVE-2026-15830 (DoS), and CVE-2026-15920 (XSS via URLField in admin).

The cause was a misconfigured eval environment with partner Irregular: test machines were given real internet access despite a system prompt stating they were isolated. The models treated the CTF target as legitimate and attacked the infrastructure of three real companies; the incident was found while reviewing 141,006 test sessions.

PayU (UPI, cards, net banking) is included as an official payment provider in Odoo 19 without a third-party module: install it from the Apps Store, set up the webhook in PayU Merchant Dashboard, and payment data flows straight into accounting.

Fixed a bug where string fields containing dates were incorrectly converted to DateTime when saved to the database. Critical integration issues with 1C and databases on Linux were also resolved, and authentication error messages were unified to resist credential stuffing.

Model Context Protocol moved from a bidirectional stateful protocol to sessionless request/response, so any request can be routed to any server instance behind a load balancer without shared state storage. Multi Round-Trip Requests were added, letting the server ask for clarification via input_required instead of keeping a stream open; Mcp-Method/Mcp-Name headers were added for gateway-level routing; authorization was tightened with issuer validation per RFC 9207 and Client ID Metadata Documents instead of Dynamic Client Registration; the extensions framework was formalized with Tasks, MCP Apps, and Enterprise Managed Authorization. All Tier 1 SDKs were updated: TypeScript, Python, Go, and C#.

The largest revision since launch. Protocol-level sessions (Mcp-Session-Id), the initialize/notifications/initialized handshake, ping, and SSE resumability were removed; cross-call state is now carried in explicit handle arguments. server/discover was added, the MRTR pattern replaced server-initiated requests, and resultType became mandatory. Roots, Sampling, Logging, and the HTTP+SSE transport are marked Deprecated (at least 12 months before removal), and Dynamic Client Registration gives way to Client ID Metadata Documents.

Opus 5 is presented as a step forward over Opus 4.8, approaching Claude Fable 5 in quality at half the price and supporting 1M-token context on a flat plan. The model is available only through Anthropic API/cloud, with no published weights and no self-hosting.

3.6 Flash - 1M-token context, knowledge cutoff March 2026, 17% fewer output tokens with stronger coding and long-context performance than 3.5 Flash. Flash-Lite is a lighter, cheaper version, and Flash Cyber is specialized for security tasks. All models are closed, API-only.

The process handling service and message store have been moved to core cluster services with auto-deployment to all nodes; internal storage and indexing in DatareonDB have been redesigned. The 1C connector can now record delete events in information registers. Default passwords have been removed - password change is required on the first sign-in to the configuration center.

New built-in analytics app (rolling out via Saleor Cloud) surfaces revenue, orders, returns, fulfillment trends and low-stock alerts directly in Dashboard, with an agent-ready data layer for AI agents to query store metrics.

Yandex was the first in CIS to receive ISO/IEC 42001 for the base YandexGPT model back in 2025; certification now covers the full Alice AI lineup (LLM, ART, VLM) - development processes and data protection passed an independent audit.

The GigaChat 3 Ultra model became available to individuals in the public API under a free freemium mode (the limit was raised to 365 million tokens by July 22, 2026). Access is cloud-based, and no weight release was announced for this specific build.

The beta feature shows Claude usage patterns (peak activity, conversation charts, task breakdown) over 1/3/6/12 months, with quiet-hour settings and break reminders. Available for Free/Pro/Max when memory is enabled.

An update to a closed model, API-only. Muse, not Llama, is established as Meta's frontier line.

A family of three models in increasing capability: Luna, Terra, Sol. Public release on July 9, 2026 after a limited preview for trusted partners starting June 26 (US export restrictions). The top-tier Sol is available only through API/ChatGPT, with closed weights. Prices were cut twice: Luna -80% and Terra -20% (July 30), Sol -20%+ for 3 months (August 21).

Studio adds a dark theme, per-language split-screen translation editing and data-quality dashboard widgets. Platform gains CDN cache-proxy support (Fastly built in) with per-asset/thumbnail invalidation via surrogate keys, and ships a Software Bill of Materials with every release.

Akeneo has added an AI agent layer (Agentic Ziggy) that enriches, validates, and orchestrates product data through natural-language commands instead of manual setup. The models are closed, access is only through Akeneo Product Cloud (SaaS), and self-hosting is not available.

A 432B MoE model (28B active) with a hybrid MLA + GatedDeltaNet architecture (linear attention): long-text generation is 4x faster, and the model is nearly twice as compact as GigaChat 3.1 Ultra while delivering similar code and math quality. Weights are open on HuggingFace (ai-sage/GigaChat3.5-432B-A28B).

Claude writes and commits code through the GitHub integration with Odoo.sh databases ("vibe coding" without manual coding); ChatGPT-5 and Gemini handle tax and expense reporting in natural language. All three models are closed-weights, available only through the vendors' cloud APIs - Odoo does not host the weights itself.

Sonnet 5 is the most agentic model in the Sonnet line: it plans, uses the browser and terminal, and works autonomously. Pricing is $2/$10 per million tokens in/out, and as of August 10, 2026 it became permanent. Weights are closed, and access is only through the Anthropic API.

A new product, not a new model: a wrapper around Opus 4.8 with integration of 60+ scientific databases for genomics and drug development. Beta on macOS/Linux for Pro/Max/Team/Enterprise; up to 50 projects receive $30,000 in compute credits.

A backport release for the 4.3 branch with several critical fixes; the main one addresses a RocksDB native memory leak in Kafka Streams state stores. A matching 4.2.1 backport was also released for the 4.2 branch.

orderGrantRefundCreate/Update mutations now accept a reasonReference on each order line, stored on OrderGrantedRefundLine, so a single refund spanning several items can carry a different reason per line.

New HOMEPAGE_WIDGETS mount point lets apps render widgets in a configurable Dashboard home grid; new App Bridge WidgetResize action added; separately shipped Anonymizer app strips personal customer data for GDPR erasure requests and dev-DB sanitizing.

Anthropic launched Claude Tag (beta, Enterprise/Team plans), where Claude joins a Slack channel as a regular participant and anyone can assign a task with an @ mention, without a separate bot for the channel. The work model is an agent in the team’s shared channel, not a developer’s personal CLI tool.

Python 3.15.0 beta 3 is out - the feature freeze has passed, and stabilization is underway before release.

The team has opened for public discussion the phase-out of native npm installation for the self-hosted instance. Docker will become the only officially supported deployment method. As of publication, no decision has been made, and the topic remains open for community comments.

The announcement of union types in C# for modeling closed data sets (wire protocols, domain models) has been a long-standing developer request. .NET 11 focuses on performance and diagnostics; ASP.NET Core and Blazor get building blocks for agentic web apps (agents, tools, skills, components). A new dotnetup tool manages SDK/runtime versions.

Each workflow run creates a root span with status, duration, and error, plus an optional child span for each node. Enabled with two env variables, exported to any OTLP collector (Jaeger, Tempo, Datadog, Honeycomb, Grafana Cloud) with no vendor lock-in.

The code of new and republished marketplace modules (2,000+ solutions) is now additionally checked by an AI analyzer, which classifies findings by CWE/OWASP Top 10 and CVSS; vulnerable code is blocked from publication until the author fixes it.

SWE-bench Pro rose to 69.2% (from 64.3% for Opus 4.7), and Terminal-Bench is up 8.5 points. 1M-token context comes at no surcharge - Anthropic keeps a flat rate for long context (Opus 4.8/4.7, Sonnet 4.6), unlike some competitors. Weights are closed.

On May 27-28, 2026, Saby (formerly the SBIS brand) held an EDI industry forum with case studies on adapting businesses to the mandatory transition to electronic waybills from September 1, 2026.

25 KIPs and 600+ commits since 4.2.0: the last ZooKeeper support has been removed from the codebase, with KRaft now the only architecture. KIP-1257 adds metrics for retention-limit usage by topic-partition. KIP-1258 adds client assertion authentication for OAuth client_credentials. The streams-scala module is marked deprecated and slated for removal in 5.0.

The monthly release added native components for AWS Bedrock and Google Gemini, enabling direct LLM provider integration in ETL jobs without custom code. It also adds transactional mode for tKafkaOutput and GA for Databricks 17.3.

n8n 2.22.0: the agent connects to some external MCP servers (Apify, Linear, Notion, and others from the official MCP registry) by selecting them from the node panel and logging in, without manually creating an MCP Client node and credential.

New Bitrix24 mode: the user describes the task in text, and an AI agent based on BitrixGPT 5.5 studies the documentation, writes the code, and assembles the app without a developer. It works only within the Bitrix24 cloud ecosystem, and model access is closed (API/subscription, weights are not published) - closed-weights tag.

581 core fixes. First-day PHP 8.5 support, PHP 8.2 support removed. Valkey 8.x (open-source Redis fork) becomes default cache/session backend. OpenSearch 3.x recommended, Elasticsearch removed entirely as search backend — merchants on 2.4.8 must migrate to OpenSearch before upgrading. Symfony 7.4 LTS.

The company valuation more than doubled in less than a year, starting at $2.5 billion. Under a commercial agreement, n8n is natively embedded in Joule Studio, the agent-building environment on SAP Business AI Platform, to automate processes across SAP systems and external software from a single interface.

Elastic Workflows (rules-based + agentic automation for security triage/response) reaches GA, replacing the need for separate SOAR tooling. Elasticsearch and Kibana get full-stack FIPS 140-3 GA ahead of the September 2026 deadline, no data migration required. Native PromQL support (technical preview) claims 30x faster queries than Prometheus itself.

Node 26 includes the Temporal API (Temporal.PlainDate, Temporal.ZonedDateTime) without flags as a replacement for the outdated mutable Date; V8 is updated to 14.6 (Chromium 146), and Undici to 8.0. Current release, entering LTS in October 2026.

Connectors for finance, marketing, CX, ITSM, security, DevOps, IT, and media systems. A continuation of the connector release series after 50 connectors in winter 2026.

The AI client through the built-in MCP server (Cloud/Enterprise/self-hosted Community) now creates new workflows and edits existing ones directly in the instance, not just executes existing ones. The intermediate representation is TypeScript, which must pass type-checking and compilation before being applied to the instance, not raw JSON. Public preview, recommended version n8n 2.18.4+.

GPT-5.5 and GPT-5.5 Pro are available in ChatGPT starting April 23, 2026 (paid plans only), and in the API starting April 24. Weights are closed: the model runs exclusively through the OpenAI cloud/API, and self-hosting is not possible.

Odoo has received Plateforme Agréée status from the French tax authorities, so receiving and sending electronic invoices and VAT e-reporting go directly through the product. From September 2026, receiving e-invoices will be mandatory for all companies in France, and issuing them as well for medium and large businesses.

The catalog model now automatically suggests attributes based on rejection logs and search trends; AI provider selection (OpenAI, Claude, etc.) has been added for content generation instead of tying it to a single model. Scaling runs on Upsun hosting, and the platform remains cloud-based.

Pimcore included as a Niche Player in Gartner's 2026 Magic Quadrant for Master Data Management Solutions (18 vendors evaluated) — its second Magic Quadrant appearance alongside the Digital Experience Platforms MQ.

The model ships with automatic filters that block forbidden high-risk cyber requests; it is available natively on Claude Platform and through Bedrock, Vertex AI, and Microsoft Foundry. Self-hosting is unavailable, and gateway routing must keep sessions in the provider's cloud.

Business users run Synapse agentic workflows for item creation, onboarding, enrichment, and syndication through natural-language chat instead of the PIM interface. Announced at Syndigo Connect.

Pimcore Studio, a React/TypeScript rewrite of the admin UI built over ~2 years, goes GA and fully replaces Admin UI Classic, whose ExtJS code is removed from the codebase. Release also introduces the Pimcore Data Spine data-architecture concept and a beta Agent SDK for embedding AI agents in the platform.

An agent inside Alice AI searches products across billions of offers online and completes the purchase directly in chat: average savings of 25% (about 6,000 RUB), with search taking 5-8 minutes. Available to 30.5 million monthly chat users; no specific agent model was named.

The first proprietary MSL model: API-only, with no published weights. A break from Llama's distribution model of downloadable open weights and self-hosting.

The Community Edition has fixed the vYYYY.Minor versioning scheme (calver) in the README instead of semver; breaking changes are allowed in any minor release, and pinning the version in composer.json is recommended.

Synapse runs AI agents across product, supplier, location, compliance, and performance data as a single digital twin, built on the multi-domain MDM core inherited from Riversand's 2021 acquisition. Human oversight of the agents is promised, not full autonomy.

Batch jobs gained support for Spark 4.0.1 (production-ready), and a native AWS Athena connector was added. Twenty-three CVEs were fixed, including vulnerabilities in Apache Avro, Log4j, Netty, and Rhino.

Through the Yandex Commerce Protocol (YCP), stores on 1C-Bitrix can show product cards and accept orders directly in Yandex Search and Alice, without going to the website; order and sales management remains with the seller.

Launched at Anthropic's first Partner Summit in Carlsbad, March 12-13, 2026. Free membership for consulting firms and systems integrators; funds go toward partner training and co-marketing. The partner-facing team is planned to grow 5x.

With Node.js 27 (October 2026), the release model changes: Alpha 6 months -> Current 6 months -> LTS 30 months. All releases become LTS, and the odd/even split is removed. The reason is minimal adoption of odd-numbered releases over 10 years and the volunteer team being overloaded by supporting four branches at once.

Fourth patch release on the 2.4.8 line, shipped alongside 2.4.9 beta availability for early compatibility testing before May GA.

Existing APIs are wrapped into an MCP server without rewriting code, configured through API Manager and run on Flex Gateway. Part of Anypoint Platform, not a separate service.

Yandex AI Studio gave businesses a builder for reasoning agents based on DeepSeek-V3.2 (third-party, not YandexGPT) for multi-step tasks such as tender analysis, incident review, and code. Caching/tool tokens were added (4x cheaper), along with secure connectivity and no request logging in the enterprise environment.

The minor release consolidates modules instead of adding features: Field Service is absorbed into Planning, Remote Work into the Employee app, and work entries are removed from Payroll and merged with absence types. Plus offline search and content caching.

New Storage Optimized (ARM) and updated CPU Optimized (ARM) profiles run on AWS Graviton4. Elastic claims up to 15% better price-performance for CPU-optimized and up to 40% for storage-optimized versus prior instance types.

A backport release for the 3.9 branch with several critical fixes, including KIP-1252, which resolves inconsistent behavior between ZooKeeper and KRaft modes. Relevant for clusters that have not yet migrated to KRaft-only 4.x.

The tool reads and traces code like a security researcher, not by pattern rules; multi-stage validation filters out false positives. On our own codebase, it found 500+ vulnerabilities missed by years of manual audits. Access is available for Enterprise/Team by request, and free and faster for open-source maintainers.

Share groups from KIP-932 are now production-ready: the RENEW acknowledgment type has been added for long processing, along with adaptive batching in the share coordinator, soft and hard limits on the number of fetched records, and lag metrics. Streams Rebalance Protocol (KIP-1071) reaches GA with a limited feature set. CLI tools now use unified arguments (--bootstrap-server everywhere).

Agent Builder (chat with Elasticsearch data, build custom AI agents) goes GA and integrates with the new (preview) Workflows engine so agents can take actions, not just answer questions. Elastic Cloud Serverless expands to 18 regions with an AWS infra upgrade (up to 35% lower search latency, 26% higher ingest throughput).

Starting with the 2026-06 release, Talend Studio requires Java 21. This affects only the desktop application; jobs and services themselves remain on Java 17 (Big Data Jobs on Java 8). Runtime (Remote/Cloud/Dynamic Engines, Job Server) will support Java 21 as an option, while Java 17 remains the default.

A maintenance release with no new features - stabilization of the portal, business processes, connectors, and the configuration import/export mechanism. Upgrading from 3.1.2 or earlier requires intermediate migration steps.

MCP server tools can return interactive HTML components, such as dashboards, forms, and multi-step wizards, rendered in an iframe sandbox directly in the chat with the model. The first extension to move from an SEP proposal (November 21, 2025) to official production-ready status.

Akeneo has expanded its MCP server to connect LLMs directly to product data while preserving governance rules, and has partnered with Stripe Agentic Commerce Suite to give AI agents access to the catalog and checkout. Digital Showroom is a showcase of enriched data for sales teams.

From January 2026 Adobe moved to: monthly isolated security fixes as needed, one major version release per year (May), and aggregated security patch rollups in May and November — replacing the prior ad-hoc patch schedule.

AI agents in Odoo now read uploaded files and answer based on their contents with period filtering; offline mode hides features unavailable without a network instead of failing with an error. Push notifications have been moved from Firebase to Odoo's own infrastructure.

Final version 8.5.1 is out, bringing all 8.3.27 features plus the 8.5 interface in beta status (web client, dark theme, new window system). UNF, Retail, and POS have been adapted at the same time; interface development will continue in version 8.5.5 throughout 2026.

Symfony 7 compatibility added alongside continued Symfony 6.4 support; Symfony 6.x removed at Platform 2026.1. Experimental MCP server exposes Pimcore-managed data to LLMs/agents; legacy ExtJS Admin UI Classic marked deprecated ahead of removal in 2026.1.

Available to Google AI Ultra subscribers by selecting Deep Think + Gemini 3 Pro in the dropdown. At launch, it scored 41.0% on Humanity's Last Exam (without tools) and 45.1% on ARC-AGI-2 (with code execution), claimed as leading results. Closed model, Google cloud only, access limited to a paid subscription.

The program on the Standoff Bug Bounty platform is open to independent researchers to find vulnerabilities in the cloud version of Bitrix24 for a reward, expanding the earlier closed security-audit partnerships.

Protocol anniversary. Experimental task-based workflows were added (SEP-1686, statuses working/input_required/completed/failed/cancelled), Dynamic Client Registration was replaced by URL-based Client ID Metadata Documents (SEP-991), URL mode elicitation for OAuth was added without passing credentials through the client (SEP-1036), tool calling inside sampling requests was added (SEP-1577), and the tool name format was standardized (SEP-986).

Standardization of UI resource declarations, their binding to tools, and bidirectional messaging between embedded UI and the host. Initially limited to text/html in a sandboxed iframe. One of the community's most requested features, building on earlier MCP-UI work and the OpenAI Apps SDK.

Major release: native pipe operator |> for left-to-right call chaining, built-in URI extension (RFC 3986 / WHATWG parsing), clone-with syntax for readonly objects, #[\NoDiscard] attribute, closures allowed in constant expressions.

Added Enhanced Content Layout Design for auto-generating rich-media item pages, AI GoPilot for extracting attributes from photos and validating data quality, and AVMT for automatically mapping brand attributes to a specific retailer's fields. The 'Is it Live' analytics were expanded to track publishing across more retailers.

.NET 10 is an LTS release supported through November 10, 2028. C# 14 adds field-backed properties, extension properties, and first-class Span<T> conversions. The platform includes Microsoft Agent Framework, which combines Semantic Kernel and AutoGen to build AI agents with MCP and A2A out of the box.

Two reasoning models (120B and 20B) for content moderation/classification based on customer policies rather than fixed OpenAI rules. Weights are open under Apache 2.0, so they can be fine-tuned and deployed in your own environment without sending data to the OpenAI cloud.

Node.js 24.11.0 entered Active LTS on October 28, 2025, with support through April 2028. AsyncLocalStorage now uses AsyncContextFrame by default, URLPattern is globally available, and npm 11 is included.

Yandex relaunched the Alice chat on the new Alice AI LLM model - a MoE architecture with hundreds of billions of parameters, trained with online RL and built-in multi-stage web search. The YandexGPT brand for the chat flagship has effectively been replaced by Alice AI; weights are closed.

Agent Builder ships as a new capability letting developers build AI agents against Elasticsearch data. DiskBBQ serves quantized vector clusters straight from disk (sub-20ms latency, ~100MB memory) for large-scale vector search. Streams auto-parses and compresses unstructured logs to surface incident root causes.

A Chromium-based browser for macOS with a built-in ChatGPT assistant in the sidebar and agent mode (ChatGPT agent), released on October 21, 2025. OpenAI shut down Atlas on August 9, 2026, moving browser agent features into ChatGPT and Codex.

Current release: the --allow-net flag gives explicit control over network access in the permission model; Web Storage (localStorage/sessionStorage) is enabled without a flag; V8 is updated to 14.1 with faster JSON.stringify and built-in Uint8Array base64/hex conversion.

Free-threading (a build without the GIL) is officially supported, per PEP 779, and the penalty for single-threaded code has been reduced to about 5-10%. The JIT compiler remains experimental (PYTHON_JIT=1). The new t-strings syntax (PEP 750) provides template strings with deferred interpolation for safe SQL/HTML construction.

Agent Builder is a visual builder for multi-agent workflows with versioning, Connector Registry is a registry of connections to data and tools, and ChatKit embeds chat agents into interfaces. Announced at DevDay on October 6, 2025 as a replacement for homemade orchestration without versioning and eval pipelines.

TinyMCE 7 editor comes back to Pimcore exclusively for paid Professional and Enterprise editions under an official OEM agreement with Tiny Technologies. Not available in the free community edition.

A video model with synchronized sound, improved physics, and multi-frame consistency; launched through the iOS app and sora.com by invite (US, Canada). Weights are closed, and access is only through the OpenAI app/API. The product was discontinued in under a year: the app and sora.com were turned off on April 26, 2026, and the API will be shut down on September 24, 2026.

A family of safety classifiers (0.6B/4B/8B) for real-time moderation of LLM traffic: Gen (instruction-following) and Stream (token-level). Apache 2.0, open weights.

AI features are natively available in every app: "Ask AI search" turns a natural-language question into Odoo filters/groupings, AI agents perform actions on request (quotes, follow-up), and fields are filled automatically by AI. ChatGPT has been added as a provider: the model is cloud-based, there are no weights, and access is only through the OpenAI API.

Queues for Kafka (share groups, cooperative consumption of one partition by multiple consumers without reassignment) is entering preview for evaluation, not production. Streams Rebalance Protocol (KIP-1071) in early access moves Kafka Streams rebalancing to the broker, following KIP-848. KIP-877 lets plugins and connectors register their own metrics.

Syndigo, Riversand's parent company, acquired 1WorldSync, including PowerReviews, adding GDSN data-pool synchronization and user-generated content such as reviews to its MDM/PIM stack. The deal was backed by Summit Partners, TJC, and Battery Ventures; the combined base has 18,000+ customers in 60 countries.

Yandex B2B Tech opened access to YandexGPT 5.1 Pro through Yandex Cloud AI Studio: the share of good answers rose from 60% to 71%, hallucinations fell by half to 16%, and the price was cut threefold to 40 kopeks per 1,000 tokens. The model is available only via API, with closed weights.

An image editing model based on Qwen-Image (native text rendering); open weights, self-hosting possible.