US Intercom Releases Customer Service AI Model Fin Apex 1.0, Achieving Higher Resolution Rate Than GPT-5.4 and Claude Sonnet 4.6

en.Wedoany.com Report on Mar 27th, The US customer service platform Intercom recently launched the AI model Fin Apex 1.0, specifically built for the customer service domain. The company stated that this model outperforms cutting-edge general-purpose models such as OpenAI's GPT-5.4 and Anthropic's Claude Sonnet 4.6 on the key metric of customer issue resolution rate.

According to the published benchmark data, Fin Apex 1.0 achieved a resolution rate of 73.1%, which is the proportion of customer issues fully handled without human intervention. In comparison, GPT-5.4 and Claude Opus 4.5 both scored 71.1%, while Claude Sonnet 4.6 scored 69.6%. Eoghan McCabe, CEO of Intercom, pointed out: "For large-scale service operations, a difference of a few percentage points translates to a significant volume of customer interactions and revenue impact."

Beyond resolution rate, the model also shows improvements in response speed and accuracy, with a response time of 3.7 seconds and a significant reduction in misinformation generation. Its operating cost is approximately one-fifth that of directly using cutting-edge models, and it is already included in Intercom's existing outcome-based pricing plan, offering a free upgrade for existing customers.

Intercom did not disclose the specific open-source foundational model or its parameter scale upon which Fin Apex 1.0 is based, only stating that the parameters are at the hundreds of billions level. The company emphasized that the model's core advantage lies in its post-training process, which involves optimizing the model using proprietary data accumulated from its customer service platform, Fin. McCabe stated: "Pre-training has become standardized; the real competition lies in post-training. Domain-specific models typically outperform general-purpose models within their specific domain."

Currently, Fin Apex 1.0 is only available through Intercom's Fin AI Agent and is not licensed separately to external parties. The company plans to expand its AI capabilities into areas such as sales and marketing. This development reflects a shift in enterprise AI applications from cost-saving to experience enhancement and may prompt a rethinking of the SaaS industry's reliance on general-purpose API models.