Intercom's new post-trained Fin Apex 1.0 beats GPT-5.4 and Claude Sonnet 4.6 at customer service resolutions

Intercom's new post-trained Fin Apex 1.0 beats GPT-5.4 and Claude Sonnet 4.6 at customer service resolutions

Intercom's Fin Apex 1.0 Surpasses GPT-5.4 and Claude Sonnet 4.6 in Customer Service Resolutions

Intercom, a 15-year-old Dublin-based customer service platform, has unveiled Fin Apex 1.0, a purpose-built AI model that outperforms leading frontier models from OpenAI and Anthropic in customer support. The model powers Intercom's existing Fin AI agent, which handles over one million customer conversations weekly. According to benchmarks shared with VentureBeat, Fin Apex 1.0 achieves a 73.1% resolution rate, surpassing GPT-5.4 and Claude Sonnet 4.6 with 71.1% and 69.6% resolution rates, respectively.

Background & Context

Intercom's move to build its own AI model is significant, especially for a legacy software company. The company's decision to invest in AI development may signal a shift in the industry, where companies are no longer relying solely on pre-trained models from established players like OpenAI and Anthropic. This development could have implications for the future of customer service and the role of AI in supporting large-scale service operations.

Impact on Swiss SMEs & Finance

The impact of Fin Apex 1.0 on Swiss SMEs and finance is multifaceted. For one, the model's ability to deliver responses in 3.7 seconds and demonstrate a 65% reduction in hallucinations could improve customer satisfaction and reduce operational costs for businesses. Additionally, the model's inclusion in Intercom's existing pricing structure could make it more accessible to smaller enterprises. However, the model's performance may also create pressure on Swiss companies to invest in AI development, potentially widening the gap between those that can afford to innovate and those that cannot.

What to Watch

As Fin Apex 1.0 gains traction, it will be interesting to see how other companies respond to Intercom's move. Will we see a surge in AI development among legacy software companies, or will established players like OpenAI and Anthropic continue to dominate the market? Additionally, Intercom's decision not to disclose the base model used for Fin Apex 1.0 raises questions about the importance of base models in AI development. Will this trend continue, and what implications will it have for the industry as a whole?

Source

Original Article: Intercom's new post-trained Fin Apex 1.0 beats GPT-5.4 and Claude Sonnet 4.6 at customer service resolutions

Published: March 26, 2026

Author: carl.franzen@venturebeat.com (Carl Franzen)

---

Disclaimer: This article is for informational purposes only and does not constitute financial advice. Consult a licensed financial advisor before making investment decisions.

Related Articles

• Anthropic’s Claude can now control your Mac, escalating the fight to build AI agents that actually do work
• Cursor's Composer 2 was secretly built on a Chinese AI model — and it exposes a deeper problem with Western open-source AI
• New MiniMax M2.7 proprietary AI model is 'self-evolving' and can perform 30-50% of reinforcement learning research workflow