Service update: WebBrain Compass restored after upstream Novita hiccups

If you ran into connection errors or timeouts with WebBrain Compass over the last couple of days, you were not alone. An upstream issue with Novita, our infrastructure provider for hosted inference, caused intermittent hiccups for Compass traffic. The underlying problem has been identified and cleared, and Compass is back to steady, normal operation.

What happened

WebBrain Compass is our hosted routing service, designed to let users test and run WebBrain out of the box without needing local GPU hardware or their own API keys. The inference backing this service is hosted on Novita's infrastructure.

Earlier this week, Novita experienced an upstream platform error that led to sporadic connection drops, degraded response times, and failed inferences for requests routed through Compass. While not a total blackout, it meant several sessions failed mid-turn or could not connect to the remote worker.

Current status

Novita resolved the incident, and our telemetry confirms that request success rates, error counts, and response latencies have returned to baseline. All WebBrain Compass endpoints are healthy.

If you bring your own API key (via OpenRouter, Anthropic, OpenAI, etc.) or run WebBrain entirely with a local model through Ollama, LM Studio, or vLLM, your setup runs locally or connects directly to your chosen provider, so you were never impacted by this incident.

Thanks for your patience while this got sorted out. If you notice any lingering issues, please let us know on GitHub or reach out in Discord.

Tags: #WebBrain #Compass #Novita #ServiceUpdate #Status