What is new in n8n 2.43?
n8n 2.43 was released on October 6, 2026. n8n's release notes list 14 feature-level changes, headed by two: Browser Use in the n8n Assistant is now available to all users, and the migration report now labels each breaking change by what it does to your workflows instead of by severity. The rest are node and editor improvements.
The source is n8n's own release notes page, which titles the entry "n8n 2.43: Browser Use now available to all users, plus 13 other features". The full code-level change list is in the GitHub release for 2.43.0. Note what n8n says on that page: publishing a feature there does not guarantee you have it yet, because some features ship behind a flag or roll out gradually.
| Change | Who it matters to |
|---|---|
| Migration report: impact labels and a "compatible of total" progress bar | Anyone planning the next major upgrade |
| Browser Use in the n8n Assistant for all users | Teams that use the Assistant to build or debug workflows |
| Azure OpenAI Chat Model node: pick a deployment from a list | Azure OpenAI users who hit DeploymentNotFound errors |
| Execute button shows the trigger icon and runs a trigger from its menu | Everyone who tests workflows in the editor |
| Supabase node: OAuth credentials, reusable for the Supabase MCP server | Supabase users |
| Databricks node: Job resource (Run, Get Job, Get Run, Get Run Output) | Data teams on Databricks |
| Crypto node: choose SHA-256 or SHA-1 for RSA-OAEP | Integrations with systems that need SHA-1 padding |
What does the new n8n migration report show?
The migration report, under Settings, now sorts every breaking-change rule into four impact labels: Upgrade blocked, Executions fail, Behavior changes and Capability removed. Blockers are listed first. A progress bar at the top shows how many of your workflows are compatible with the target version, as "X of Y compatible".
The four impact labels, in fix order
Before 2.43, the report used low, medium and critical severities. n8n says each rule has been reclassified "based on its actual runtime effect". That is more useful, because "critical" did not tell you whether a workflow would stop, change its output quietly, or simply lose a feature you never used.
| Label | What it means for you | What to do |
|---|---|---|
| Upgrade blocked | The upgrade will not go through until this is fixed | Fix these first; nothing else matters until they are gone |
| Executions fail | The workflow will error after the upgrade | Fix before upgrading, starting with workflows that run on a schedule |
| Behavior changes | The workflow runs, but may return different data | Run a test with real data before and after, and compare the output |
| Capability removed | A feature or node option goes away | Check whether you use it; plan a replacement if you do |
The "Behavior changes" group is the one we would read most carefully. A workflow that fails is noticed the same day. A workflow that keeps running and returns slightly different data can feed wrong values into a CRM for weeks. That risk is highest in workflows that sync contacts into GoHighLevel or another CRM, like the ones in our guide to GoHighLevel integrations with n8n and webhooks.
Does the migration report help with the n8n 3.0 upgrade?
Yes. The report checks your workflows against the breaking-change rules for the target version, which for most teams is n8n 3.0. It turns the list of removed nodes and changed defaults into a list of your own workflows that need work, with the blockers first.
When we read n8n's breaking-changes page on October 5, 2026, it described 3.0 as scheduled for October 2026 and listed removed nodes such as Function, Cron, Interval and Read Binary File, plus the move to Docker-only self-hosting. Our n8n 3.0 breaking changes note has the full list. The 2.43 report is the fastest way to see which of those affect you. One related change in the 2.43.0 GitHub release: legacy Cron node schedules now run on n8n's durable scheduler.
What is Browser Use in the n8n Assistant?
Browser Use lets the n8n Assistant work in a web browser while it helps you, for example to read a page or check something an API does not expose. From 2.43 it is available to all users instead of being an experiment. Instance admins control it with the setting N8N_INSTANCE_AI_BROWSER_USE_ENABLED, which is on by default.
Two details from the release note. It is not available on phones and tablets, including iPads in desktop mode. And the experimental "Set up automatically" option for credentials is now off by default for everyone.
If you run a self-hosted instance with client data, decide on purpose whether the Assistant should browse. The setting is on unless you turn it off, so check it before your team starts using it on live work.
What else is in the 2.43.0 code release?
The GitHub release for 2.43.0 also lists changes that are not in the feature notes: an opt-in budget guardrail and usage settings for agent runs, a block on publishing workflows that have unconnected required inputs, and a stop on Execute and Publish when a node uses a credential that is not yours. Treat these as coming, not confirmed, until n8n documents them.
All three are useful. Budget limits on agents stop a looping agent from running up a model bill. The publishing check catches a half-finished workflow before it goes live. The credential rule matters on shared instances, where someone could otherwise run a workflow with a colleague's connection. Because n8n's docs do not describe them yet, check your own instance after updating before you rely on any of them. For how n8n's agent feature works today, see our note on the n8n Agents launch.
Should you upgrade to n8n 2.43 now?
If you are on n8n Cloud, updates come through n8n and you can check your version in the admin panel. If you self-host, test 2.43 on a copy first. On October 12, 2026 n8n's stable tag on GitHub pointed to 2.42.7, and 2.43 builds were still marked as pre-releases.
- Open the migration report on your current version and export the list of flagged workflows.
- Test 2.43 on a staging copy with your real workflows, not a blank instance.
- Run your scheduled workflows by hand once, because a stopped schedule does not show an error on screen.
- Decide on Browser Use and the Assistant settings before the team starts using them.
- Upgrade production when the stable channel moves to 2.43, and keep a backup of the database and your workflow exports.
If your team relies on n8n for lead routing, CRM sync or AI agents, we can run the report, fix what it flags and handle the upgrade. See our n8n automation service, compare costs on our n8n and Zapier pricing research or run your own numbers in the Zapier vs n8n cost calculator. The Ultra Cryo & Recovery case study shows n8n running next to GoHighLevel for a Florida clinic.
Questions about n8n 2.43
When was n8n 2.43 released?
n8n's release notes give October 6, 2026 for version 2.43, and the GitHub release for 2.43.0 is dated the same day. Patch releases followed, up to 2.43.4 on October 12, while the stable tag stayed on the 2.42 line.
Is n8n 2.43 a stable release?
Not yet, as of October 12, 2026. The 2.43 builds on GitHub are marked as pre-releases and the stable tag points to 2.42.7. n8n moves the stable tag to a newer line after testing, so check the tag before you upgrade a production instance.
Where is the migration report in n8n?
The release note places it under Settings. It lists every breaking-change rule that affects your workflows, grouped by the four impact labels, with a progress bar showing how many workflows are compatible with the target version.
Sources
- n8n Docs, Release notes, entry "n8n 2.43", released October 6, 2026, read October 12, 2026.
- GitHub, n8n@2.43.0 release, October 6, 2026, and the releases list (stable tag on 2.42.7), read October 12, 2026.
