# Boson AI’s Higgs Realtime scores higher than GPT and Gemini in interruption recovery

Digest AI · Agents & Tools · published 2026-11-03T17:39:49Z

Canonical: https://digestai.news/story/boson-ais-higgs-realtime-scores-higher-than-gpt-and-gemini-in-interrup

## Summary

Boson AI’s **Higgs Realtime**, an audio-native speech-to-speech model, outperformed competitors in handling conversational interruptions. In Boson’s **IHBench** benchmark, it scored **71.2** on recovery quality, ahead of **GPT Realtime 1.5 (65.5)**, **GPT Realtime 2 (62.4)**, and **Gemini 3.1 Flash Live (61.1)**. The model’s strength lies in real-time processing, where it adapts to mid-conversation changes—like corrections or interruptions—without requiring restarts, a gap Boson highlights in open-weight models.

Boson’s approach contrasts with traditional pipelines (speech recognition → language model → text-to-speech) by treating audio as a unified input/output stream. Co-founder **Alex Smola**, who previously led AI initiatives at **AWS** and researched at **Carnegie Mellon University**, emphasizes **inference economics** and **multilingual consistency** as key differentiators. The company’s **ProactBench** research explores proactive assistance (e.g., suggesting remote appointments if a user mentions mobility limitations), but Smola stresses balancing anticipation with user control. Boson claims its system handles **thousands of concurrent calls** for a global insurance company at **one-third the cost of human agents**, with conversation quality improving from **30 to 80** over five months—far exceeding **GPT-4o Realtime’s 55** on the same metric.

## Key points

- Boson AI’s **Higgs Realtime** scores **71.2** on interruption recovery, ahead of GPT and Gemini models in IHBench benchmark
- Model supports **100+ languages**, including mid-conversation language switching, with real-time audio processing for natural interactions
- Boson’s system handles **live customer calls** for an insurance company at **1/3 human agent cost**, improving quality from **30 to 80** over five months

## Why it matters

Higgs Realtime’s superior handling of interruptions and multilingual adaptability could accelerate enterprise adoption of voice agents, addressing a critical gap in today’s open models. Its cost efficiency and real-world performance—proven in live contact centers—may redefine expectations for AI assistants in customer service.

## Sources

1. [Alex Smola, Co-Founder of Boson AI](https://unite.ai/alex-smola-co-founder-of-boson-ai-interview-series) (Unite.AI, 2026-11-03)

## Cite

Digest AI, "Boson AI’s Higgs Realtime scores higher than GPT and Gemini in interruption recovery", 3 November 2026, https://digestai.news/story/boson-ais-higgs-realtime-scores-higher-than-gpt-and-gemini-in-interrup

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/boson-ais-higgs-realtime-scores-higher-than-gpt-and-gemini-in-interrup.json