v6.34.0

Summary

This release moves the node from openai SDK ^6.32.0 to ^6.34.0 and closes the remaining Conversations parity gap.

Highlights

• Updated the package dependency to openai@^6.34.0
• Aligned Create Conversation Item with the upstream OpenAI Conversations contract
• Added assistant-message phase support for Conversations items, including commentary and final_answer
• Added a new importable Conversations example flow
• Updated README and in-editor help to match the shipped Conversations contract
• Added BDD feature specs and stronger parity tests for the Conversations surface
• Added verification coverage for Responses computer keypress actions that use keys arrays

Important v6.34.0 Update

Create Conversation Item now expects msg.payload.items as an array.

Older flows that send a singular msg.payload.item object need to be updated for this release.

Notes

This release is focused on exact contract alignment, documentation clarity, and stronger parity evidence around the shipped Node-RED surface.

Full Changelog: v6.32.0...v6.34.0

v6.32.0

This release brings our Node-RED OpenAI API node up from OpenAI Node SDK v6.29.0 alignment to v6.32.0 alignment.

What changed:

• Bumped our package version to 6.32.0 and aligned the openai dependency to ^6.32.0.
• Updated Batch coverage so our docs and tests now explicitly reflect support for the /v1/videos batch endpoint.
• Brought Responses coverage up to date with current gpt-5.4 model ids, including gpt-5.4-mini, gpt-5.4-nano, and dated variants.
• Added explicit Responses helper coverage for parse and stream.
• Verified and documented deferred MCP tool loading via defer_loading.
• Expanded Vector Store coverage with direct vector store search support.
• Added full ComparisonFilter parity for Vector Store search, including the in and nin operators.
• Closed remaining Vector Store family gaps around file and file-batch helper methods, including create-and-poll and upload-and-poll flows.
• Added coverage for vector store file attribute updates and parsed file-content retrieval.
• Refreshed editor labels, help text, examples, README notes, and regression tests so the shipped node behavior is clear and traceable.

In short: this release is mainly a parity and completeness update. It brings the node into line with the current OpenAI Node SDK surfaces we track, tightens up docs and examples, and adds regression coverage around the newly aligned behavior.

---

Full Changelog: v6.29.0...v6.32.0

v6.29.0

What’s New

This release brings the Node-RED OpenAI API node up to date with the latest OpenAI Node SDK through v6.29.0.

That includes the bigger v6.28.0 parity work as well as the smaller v6.29.0 follow-up.

On the video side, the node now covers the newer OpenAI video methods for character creation, character retrieval, video edit, and video extension, alongside the existing video flow support already in the project. The editor surface, in-node help, examples, and README were updated so the video capabilities are easier to find and understand.

On the Responses side, the WebSocket path was tightened for custom API bases. If you are using a compatible provider and your base URL includes query parameters, the WebSocket URL is now built correctly without changing the existing Node-RED connect, send, and close workflow.

This release also improves the shipped examples. Import-ready video coverage was added, and the example flows no longer come bundled with a prewired Service Host config node. That means imported flows arrive in a cleaner state and are ready for you to configure with your own host settings.

Finally, this release includes the v6.29.0 custom voice alignment work. Speech-related documentation now reflects the current OpenAI voice support more accurately, including both built-in voices and saved custom voice ids where supported. The runtime already handled that cleanly, so this part of the update was mainly about making sure the docs and tests matched the real behavior.

In Short

This release updates the project through OpenAI Node SDK v6.29.0, including:

• expanded video support
• cleaner Responses WebSocket handling for custom API bases
• better examples and README coverage
• updated speech voice documentation and regression coverage

Full Changelog: v6.27.0...v6.29.0

v6.27.0

v6.27.0 Release Details

This release brings the node up through the OpenAI Node SDK 6.27.0, expands the Responses and Realtime surfaces, adds persistent Responses websocket support, and refreshes the docs and examples to better match the current AI workflow landscape.

Highlights

• OpenAI Node SDK target updated to ^6.27.0
• Responses API alignment for:
  • phase
  • prompt_cache_key
  • tool_search
  • GA computer-use payloads
• New persistent Responses websocket method context:
  • connect
  • send
  • close
• Realtime docs and examples updated for the correct client-secret contract
• Realtime model guidance expanded for gpt-realtime-1.5 and gpt-audio-1.5
• New example flows for:
  • Responses phase
  • Responses tool_search
  • Responses computer use
  • Responses websocket lifecycle
  • Realtime client secrets
• README refreshed with a more modern, workflow-oriented project overview

Notes

• Added explicit ws dependency for Responses websocket support
• Websocket auth routing supports default bearer auth, custom auth headers, and query-string auth
• Long-lived websocket connections are cleaned up on node close/redeploy

Verification

npm run build
npm run test

- 29 tests passed
- 0 failed

Full Changelog: v6.22.0...v6.27.0

v6.22.0

• Added environment variable support for service host configuration values.
• OpenAI Node SDK upgraded from 4.103.0 to 6.22.0.
• Added responses.cancel.
• Added responses.compact.
• Added responses.input_tokens counting support.
• Added Conversations API support:
  • create/retrieve/modify/delete conversation
  • create/retrieve/list/delete conversation items
• Added Containers and Container Files support.
• Added MCP tool use example flow at examples/responses/mcp.json.
• Added Skills API support:
  • list/create/retrieve/modify/delete skills
  • retrieve skill content
  • list/create/retrieve/delete skill versions
  • retrieve skill version content
• Added Evals API support:
  • list/create/retrieve/modify/delete evals
  • list/create/retrieve/cancel/delete eval runs
  • list/retrieve eval run output items
• Added Realtime API support:
  • create client secret
  • accept/hangup/refer/reject SIP calls
• Added Videos API support:
  • list/create/retrieve/delete videos
  • download video content
  • remix videos
• Added Webhooks utility support:
  • unwrap signed webhook payloads
  • verify webhook signatures
• Service Host auth routing now applies Auth Header configuration at request time.

v1.103.0-patch.1

Enhancements

• Upgraded the OpenAI API Library dependency from v4.89.0 to v4.103.0

Notable Features & Changes

• Added missing delete method to the messages API.
• Full Model Context Protocol (MCP) tool support for the responses API.
• Added support for the new container endpoint.
• Added support for the new containerFiles endpoint.
• Added a simple MCP tool use example flow to the examples directory. See: MCP Example.
• Refactored code to greatly improve maintainability and stability.

Full Changelog: v1.103.0...v1.103.0-patch.1

v1.103.0

• Full Model Context Protocol (MCP) tool support for the responses API.
• Added support for the new container endpoint.
• Added support for the new containerFiles endpoint.
• Added a simple MCP tool use example flow to the examples directory. See: MCP Example.
• Refactored code to greatly improve maintainability and stability.

Full Changelog: v1.89.0...v1.103.0

v1.89.0

• Ehancements:
  • Upgraded the OpenAI API Library dependency from v4.87.3 to v4.89.0
• Features:
  • New audio helpers
  • New API models for TTS, STT, + new audio features for Realtime

Full Changelog: v1.87.3...v1.89.0

v1.87.3

• Ehancement: Upgraded the OpenAI API Library dependency from v4.85.4 to v4.87.3.

• Features: Implemented Responses API capabilities.

  • OpenAI's most advanced interface for generating model responses.
  • Create stateful interactions with the model, using the output of previous responses as input and more.
  • 🔥 Native agentic web search capabilities:

  msg.payload = {
    "model": "gpt-4o-mini",
    "tools": [{ type: "web_search_preview" }],
    "input": "What was a positive news story from today?"
  }

  • 🔥 Native agentic computer use capabilities.
  • 🔥 Reasoning model control properties:

  msg.payload = {
    "model": "o3-mini",
    "input": "How much wood would a woodchuck chuck?",
    "reasoning": {
      "effort": "high"
    }
  }

Full Changelog: v1.85.4...v1.87.3

v1.85.4

• Ehancement: Upgraded the OpenAI API Library dependency from v4.85.1 to v4.85.4.
• Enhancement: Implemented Chat Completion storage methods.

Full Changelog: v1.85.1...v1.85.4