Product Updates - October 2026: ZeroRuntime Simulations, Inference Model Additions, and Reliable PubSub
Blog post from Video SDK
VideoSDK’s October 2026 updates introduce simulation tools for testing ZeroRuntime voice agents through either a dashboard or Runtime SDK workflow, reducing reliance on live calls during development. ZeroRuntime Python SDK v0.1.9 adds managed OpenAI and Anthropic inference for language, speech, and real-time models, alongside updated provider defaults, prompt caching, live console logs, and more detailed session traces for debugging agent behavior. Conversation features now include configurable warm transfers with separate consultation records, custom hold and failure messages, interruption phrases, and asynchronous knowledge-base retrieval with observability support. JavaScript SDK v1.2.1 and React SDK v1.2.0 improve PubSub reliability by adding main-thread fallback when Web Workers fail, isolating reconnection issues, and returning explicit errors for operations attempted after meetings close. Additional improvements cover WebView image capture, webcam handling, Deepgram Flux text-to-speech, streaming tool calls, OpenAI speech integrations, and interruption audio behavior, while VideoSDK has also become available through Microsoft, Google Cloud, and AWS marketplaces.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| Real-time | 4 | 4,128 | 1,085 | 233 | -15% |
| Voice AI | 4 | 2,726 | 252 | 54 | -20% |
| Observability | 2 | 3,023 | 723 | 183 | +9% |
| AI Agents | 1 | 6,890 | 1,628 | 266 | +23% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.