Data and AI releases in February 2026

Smart Diagnosis, IPv6 Support, Agent Skill 2.0

AI agents keep getting better at writing code. The tricky part is everything that comes after: deploying, debugging, and keeping things running. This week, we're making that easier on both sides. Smart Diagnosis brings an AI agent to your failed deployments, reading through logs and config to tell you what broke.

reasoning_effort="none" support for GLM 4.7 zai-glm-4.7 now accepts the reasoning_effort parameter

reasoning_effort="none" support for GLM 4.7 zai-glm-4.7 now accepts the reasoning_effort parameter. Reasoning is enabled by default , set reasoning_effort="none" to disable it for simple tasks where verbose reasoning blocks are unnecessary. The disable_reasoning parameter continues to be supported but will be deprecated soon.

Prefect 3.6.20 - new cli, who dis?

Enhancements ➕➕ Migrate CLI from typer to cyclopts, remove toggle and delegation machinery Route ProcessPoolTaskRunner subprocess events and logs through a single parent queue, improving per-task performance by ~28% in log-heavy workloads Add configurable subprocess message processors to ProcessPoolTaskRunner for custom event/log handling in the parent...

Accelerated checkouts now support multiple product variants on the product page

Accelerated checkout now supports adding multiple product variants from the product page. Theme devs can include additional items alongside the main product (for example, a TV and its warranty), allowing customers to purchase products and their add-ons using accelerated checkout Learn more in our dev docs .

Customize dashboard templates with Grafana Assistant

We’ve integrated Grafana Assistant into dashboard templates, so you can now ask the Assistant to help you with the process. Templates are a powerful way to create dashboards quickly, but they don’t always match your real-world metric names or data sources.

Automatic PHP bot scan blocking now live on all plans

Netlify now automatically blocks bot scans targeting PHP paths across all plans , no configuration required. Previously, these bots generated noise in Observability logs and metrics. They showed up without a User-Agent header. Netlify now blocks them at the edge.

Postgres version updates, TimescaleDB on PG 18, Claimable Postgres agent skill, and more release - Feb 27, 2026

Postgres version updates. We updated supported Postgres versions to 14.21, 15.16, 16.12, 17.8, and 18.2, respectively. When a new minor version is available on ...

Implementation

Send a Configure control message over your Flux WebSocket connection: { "type": "Configure", "thresholds": { "eot_threshold": 0.8, "eot_timeout_ms": 5000 }, "keyterms": ["product_name", "feature_name"] } You'll receive a ConfigureSuccess response confirming the update, and changes apply immediately in the audio stream order.

- Write support for the MCP server is now live: Users can now perform write operations via the MCP Server...

- Write support for the MCP server is now live: Users can now perform write operations via the MCP Server using the new query_rw tool. This enables programmatic data modification workflows in addition to read queries. Refer to the documentation and setup guide for read-only access for more details.

arize-phoenix-evals: v2.11.0

2.11.0 (2026-02-27) Features evals: add conciseness classification evaluator (d43c4ee) evals: add refusal evaluator (45a88fc) evals: Add utility to convert Phoenix prompts to evals templates (ded46ba) Bug Fixes evals: handle litellm 1.81.16 no longer auto-converting usage=None to Usage(0,0,0) (ad516c8) update slack invitation link (316833e) This PR was...

cline 3.68.0

Added Add dynamic Cline provider model fetching from Cline endpoint Add additional Markdown formatting in CLI Add focus indicator on action buttons in extension Fixed Clear all OCA secrets on auth refresh failure to prevent re-auth loops Resolve "Could not find the file context" error in Explain Changes Use JSON_SCHEMA for yaml.load to prevent unsafe...

Renaming Unblocking Browser

We’re making a change to our product naming: Unblocking Browser is now Headless Browser. This is a name change only and all core features, functions, and integrations will remain the same as per pre-existing workflows.

Actions - Modules - EA

We are excited to announce that __Actions Modules__ is now available in [__Early Access__](https://auth0.com/docs/troubleshoot/product-lifecycle/product-release-stages#early-access "Early Access"). This feature allows you to create, manage, and share reusable code across different Actions within your Auth0 Tenant.

Codex to Figma

Use the Figma MCP server with Codex to capture full screens or specific elements of your applications and convert them to editable design layers in Figma.

Deeplink to AI coding tools

Starting an issue used to mean manually creating a feature branch. Now it means assembling the right context so your coding agent can take a first pass. We've made this much easier to do in Linear.

Cache - Asynchronous stale-while-revalidate

Cloudflare's stale-while-revalidate support is now fully asynchronous. Previously, the first request for a stale (expired) asset in cache had to wait for an origin response, after which that visitor received a REVALIDATED or EXPIRED status.

arize-phoenix: v13.4.0

13.4.0 (2026-02-26) Features ability to show sql in stdout (f21d681) add nav counters for projects, datasets, evaluators, and prompts (77780fc) add PHOENIX_MIGRATE_INDEX_CONCURRENTLY flag for rolling deployments (058e0ca) add px api graphql command to Phoenix CLI (67e8111) app: add BashBlock for standalone shell snippets (6460df9) app: migrate bundler from...

NeMo NVIDIA Neural Modules 2.7.0

Highlights Speech Adds Per-Stream Phrase Boosting in ASR Decoding (Transducers) #15125 Adds support for streaming speech translation #15132 Released new model nemotron-speech-streaming-en-0.6b that performs English Streaming ASR Released new TTS model magpie_tts_multilingual_357m for multilingual Text-to-Speech Starting with the next release, NeMo 2.8.0...

Expo SDK 55

Learn about the changes and how to upgrade

Label management: a central place for shared labels across alerts, incidents, and SLOs

You can now create and manage labels from a single location in Grafana Cloud. Label management , available under Alerts and IRM in the Grafana main menu, provides a central view of the labels you use across your alerting, incident response, and operational workflows. Labels are shared across Alerting, IRM, and SLOs.

Everything we covered at Release Notes Feb '26 livestream

In our latest Release Notes livestream, we covered the latest updates across our products, a live Q&A with Figma’s Chief Product Officer, and more, including a new way to work in Figma. Move from concept to completion using AI image tools, Make, and interactive prototypes.

UpdateThink: Replace Think Provider Mid-Conversation

You can now dynamically replace the entire Think provider configuration during a live Voice Agent conversation using the new UpdateThink client message. This allows you to switch to a different LLM provider, change the model, set a completely new prompt, and reconfigure functions , all in a single message.

Notion 3.3: Custom Agents

Prefect 3.6.19 - This release sucks in the best way possible

This release introduces an optional database vacuum service that periodically deletes old flow runs and orphaned resources (logs, artifacts, artifact collections) to keep your database lean. It's disabled by default , enable it with PREFECT_SERVER_SERVICES_DB_VACUUM_ENABLED=true.

Connect FigJam to Notion

FigJam is now available in Notion as a pre-configured MCP integration.

GPT-5.3-Codex now available in AI Gateway

OpenAI's GPT-5.3-Codex model is now available through Netlify's AI Gateway with zero configuration required. Use the OpenAI SDK directly in your Netlify Functions without managing API keys or authentication. The AI Gateway handles everything automatically.

cline 3.67.0

Added Add support for skills and optional modelId in subagent configuration Add AgentConfigLoader for file-based agent configs Add Responses API support for OpenAI native provider Preconnect websocket to reduce response latency Fetch featured models from backend with local fallback Add /q command to quit CLI Add MCP enterprise configuration details Pull...

AI hosting controls (Governance+)

Governance+ org admins can now turn on AI hosting controls so AI requests stay within Figma’s AWS environment.

Agents, R2, Containers - Backup and restore API for Sandbox SDK

Sandboxes now support createBackup() and restoreBackup() methods for creating and restoring point-in-time snapshots of directories. This allows you to restore environments quickly. For instance, in order to develop in a sandbox, you may need to include a user's codebase and run a build step.

February Release Spotlight

OpenRouter has added model benchmarks to its pages, introduced a free model router, and improved cost transparency across providers.

truss 0.14.0

- truss push: Make published deployments the default + --watch option More graceful validation for TrussTRTLLMBuildConfiguration Release 0.14.0

unstructured 0.21.0

0.21.0 Fixes - Replace NLTK with spaCy to remediate CVE-2025-14009: NLTK's downloader uses zipfile.extractall() without path validation, enabling RCE via malicious packages (CVSS 10.0, no patch available). spaCy models install as pip packages, eliminating the vulnerable downloader entirely.

New rejection reason codes and merchant message added to `verificationSessionReject` mutation

The verificationSessionReject mutation in the Payments Apps API now supports two new rejection reason codes and an optional merchant-facing message. The new enum values on VerificationSessionStateReason include: RESOURCE_NOT_FOUND : The payment method could not be found in the payment provider's system.

Agents, Workers - @cloudflare/codemode v0.1.0: a new runtime agnostic modular architecture

The @cloudflare/codemode ↗ package has been rewritten into a modular, runtime-agnostic SDK. Code Mode ↗ enables LLMs to write and execute code that orchestrates your tools, instead of calling them one at a time. This can (and does) yield significant token savings, reduces context window pressure and improves overall model performance on a task.

N-Day LTV in Retention (Custom Revenue + Change Over Time)

What it is: N-Day LTV brings cumulative revenue measurement to Retention Analysis, so you can track how much value a cohort generates across fixed day windows: Day 1, Day 7, Day 14, Day 30. To use it, set a start event, pick a revenue return event, and toggle N-Day LTV under the Return Event controls.

Official n8n Node for Resend

You can now send emails, manage contacts, and trigger workflows from n8n.

New Make connectors

From design system documentation and PRDs to user research and feedback, Make can now pull in context from across your product ecosystem.

DDoS protection, Railway domains, better canvas state

We've partnered with [Fastly](https://www.fastly.com/) to bring DDoS protection to every public service on Railway. All traffic routes through their global network (100+ points of presence) where it's filtered before it reaches your infrastructure. Nothing to configure, nothing to enable. It's on by default.