### Hotfix on top of 0.90.4
* chore(docker): upgrade Node.js to 24.22.1
Fixes the [Node.js October 2026 security vulnerabilities](https://nodejs.org/en/blog/vulnerability/october-2026-security-releases).
### Hotfix on top of 0.89.0
* chore(docker): upgrade Node.js to 24.22.1
Fixes the [Node.js October 2026 security vulnerabilities](https://nodejs.org/en/blog/vulnerability/october-2026-security-releases).
## What's Changed
### Features
- Support conversational flow context compaction.
### Bug Fixes
- Update vulnerable audit dependencies.
- Stop sending 'stop' to GPT-6 and newer models.
- Run the provider listing off the loop in `acleanup_provider_files`.
- Emit `LLMGuardrailCompletedEvent` for a malformed guardrail result.
- Name the renamed integration token when it is missing.
- Fix path traversal in `output_file` via template interpolation.
- Patch deserialization, SQL injection, and path traversal vulnerabilities.
- Restore configured logging levels after suppression.
- Drop checkpoint resume state when copying a crew.
- Honor an injected `aioboto3` session in `BedrockFileUploader`.
- Use the right task slice with custom templates.
- Guard against storage listing failures in encoding flow analysis.
- Enforce guardrail result/error exclusivity with model validator.
- Filter categories server-side in Qdrant edge search.
### Documentation
- Use HTTPS for CrewAI AMP link in CLI conce
## What's Changed
### Features
* feat(ai-gateway): record model, key owner, byte and upstream error facts on generations by @hassiebp in https://github.com/langfuse/langfuse/pull/19161
* feat(ai-gateway): product analytics for setup, credentials, invites, keys and model access by @wochinge in https://github.com/langfuse/langfuse/pull/19379
* feat(ai-gateway): restore empty-state placeholders for providers, models and gateway API keys by @wochinge in https://github.com/langfuse/langfuse/pull/19391
* feat(ee)!: add self-hosted government plan for lf_gov_ license keys by @Steffen911 in https://github.com/langfuse/langfuse/pull/19316
* feat(ai-gateway): pick team recipients with member pills instead of a notify-all toggle by @wochinge in https://github.com/langfuse/langfuse/pull/19393
* feat(topics): add Show unassigned toggle to the Topics actions menu by @nimarb in https://github.com/langfuse/langfuse/pull/19423
### Fixes
* fix(web): Topics trace list loads rows as you s
## What's Changed
* fix(web): check that an invite's project belongs to the organization (v3 backport of #19404) by @Steffen911 in https://github.com/langfuse/langfuse/pull/19408
* fix(annotation-queues): validate queue when adding items (v3 backport of #19407) by @Steffen911 in https://github.com/langfuse/langfuse/pull/19409
* fix(automations): mask and encrypt legacy webhook headers (v3 backport of #19207) by @Steffen911 in https://github.com/langfuse/langfuse/pull/19421
* fix(security): require secrets again when webhook or PostHog destination changes (v3 backport of #19261) by @Steffen911 in https://github.com/langfuse/langfuse/pull/19422
**Full Changelog**: https://github.com/langfuse/langfuse/compare/v3.225.14...v3.225.15
## 1.34.0 (2026-10-27)
<!-- Release notes generated using configuration in .github/release.yml at main -->
## What's Changed
### Features
* feat(verify): stand in for DigitalOcean so Managed Agents can be verified by @enyst in https://github.com/OpenHands/OpenHands/pull/18867
* feat(canvas): per-conversation Docker runtime and Settings → Runtime page by @VascoSch92 in https://github.com/OpenHands/OpenHands/pull/18436
* feat(automations): filter and edit automation tags by @VascoSch92 in https://github.com/OpenHands/OpenHands/pull/18767
* feat(conversation): label the share toggle "Share with Org users" in team orgs (OHE-3765) by @hieptl in https://github.com/OpenHands/OpenHands/pull/18986
* feat(sidebar): start a new conversation in the row's sandbox (OHE-3689) by @hieptl in https://github.com/OpenHands/OpenHands/pull/18912
### Bug Fixes
* fix: pass session API key to static-server via env instead of argv by @putramkti in https://github.com/OpenHands/OpenHands/pull/16420
* fix(chat):
## Features
- feat(db): add manual Trigger.dev task for the account token sealing backfill (#9393)
- feat(tailscale): expand and validate API operations (#9385)
## Improvements
- improvement(changelog): lead with knowledge base portability (#9377)
## Bug Fixes
- fix(ci): only migrate an environment from its own branch (#9395)
- fix(ci): queue migrate runs per environment instead of letting them overlap (#9394)
- fix(credentials): send a conflicting OAuth reconnect back with a reason instead of a 500 (#9391)
- fix(chat): normalize markdown list and heading spacing (#9392)
- fix(isolated-vm): drain worker host deploys without failing in-flight executions (#9389)
- fix(deps): backport zod's leading collection size check so oversized bodies stop validating every element (#9388)
- fix(library): consolidate duplicate sections and FAQs added by repeated refreshes (#9390)
- fix(ci): find the last IndexNow submission through its check run, not the run's job list (#9384)
- fix(deps): update
### Hotfix on top of 0.88.4-hotfix.3
* chore(docker): upgrade Node.js to 24.22.1
Fixes the [Node.js October 2026 security vulnerabilities](https://nodejs.org/en/blog/vulnerability/october-2026-security-releases).
### Hotfix on top of 0.91.4
* chore(docker): upgrade Node.js to 24.22.1
Fixes the [Node.js October 2026 security vulnerabilities](https://nodejs.org/en/blog/vulnerability/october-2026-security-releases).
## Liquid AI LFM3
[LFM3](https://ollama.com/library/lfm3) is Liquid AI's new family of small models built for local agents, tool calling, and structured output, on Liquid's new tri-hybrid architecture. The family includes:
- [lfm3](https://ollama.com/library/lfm3): a 1.7B text model with a 256K-token context window
- [lfm3-vl](https://ollama.com/library/lfm3-vl): adds image understanding
- [lfm3-audio](https://ollama.com/library/lfm3-audio): speech recognition, translation, and text-to-speech in 18 languages
- [lfm3-omni](https://ollama.com/library/lfm3-omni): text, images, and audio in one model
```sh
ollama run lfm3
```
## d1 decision models
Liquid AI's d1 models now come in a 2B size, built on LFM3. [d1](https://ollama.com/library/d1) makes decisions about text and images, and [d1-omni](https://ollama.com/library/d1-omni) adds audio.
```sh
ollama run d1:2b "I was charged twice for order 4417." --ask "Is this a billing issue?"
```
Note: lfm3 and d1 2b are
## What's Changed
* security: rebuild to pull node.js patches
* test(worker): skip Azure LLM connection tests until CI key is replaced (v3 backport of #17924) by @Steffen911 in https://github.com/langfuse/langfuse/pull/19386
**Full Changelog**: https://github.com/langfuse/langfuse/compare/v3.225.13...v3.225.14