Intercom’s Fin Apex 1.0 Outperforms OpenAI and Anthropic in Customer Support AI
Intercom, a customer service platform founded in 2011, is challenging the dominance of large AI models from OpenAI and Anthropic with its new, purpose-built AI, Fin Apex 1.0. Announced on Thursday, March 26, 2026, the company claims its model surpasses GPT-5.4 and Claude Opus 4.5 in key customer support metrics.
Fin Apex 1.0 currently powers Intercom’s Fin AI agent, which handles over two million customer conversations each week. According to benchmarks shared with VentureBeat, Fin Apex 1.0 achieves a 73.1% resolution rate – the percentage of customer issues resolved without human intervention – compared to 71.1% for both GPT-5.4 and Claude Opus 4.5 and 69.6% for Claude Sonnet 4.6.1
Speed, Accuracy, and Cost Efficiency
Beyond resolution rates, Fin Apex 1.0 demonstrates improvements in speed and accuracy. The model responds in 3.7 seconds, 0.6 seconds faster than its competitors, and exhibits a 65% reduction in hallucinations compared to Claude Sonnet 4.6.1 Intercom also states that Fin Apex 1.0 operates at roughly one-fifth the cost of using frontier models directly and is included in the company’s existing per-outcome pricing structure.
The Mystery of the Base Model
While Intercom touts the performance of Fin Apex 1.0, the company has declined to specify the base model upon which it was built, citing competitive reasons and plans to switch base models over time.1 A spokesperson confirmed the model is “in the size of hundreds of billions of parameters.”1
This approach follows a lesson learned from the backlash faced by AI coding startup Cursor, which was criticized for not fully disclosing its use of fine-tuned open-weights models.1 Intercom maintains transparency by acknowledging the use of an open-weights model, but remains tight-lipped about the specific one.
Post-Training as a Key Differentiator
Intercom CEO Eoghan McCabe argues that pre-training is becoming commoditized, and the real competitive advantage lies in post-training. He emphasizes the importance of proprietary data and reinforcement learning systems grounded in real-world resolution outcomes.1
The company post-trained its model using years of customer service data collected through Fin, focusing on factors like appropriate tone, judgment calls, and conversational structure. This approach aims to create an AI agent that not only answers questions but also understands when an issue is truly resolved.
Financial Impact and Future Outlook
Intercom’s AI-first strategy appears to be yielding positive results. Fin is approaching $100 million in annual recurring revenue and is growing at 3.5x, making it the fastest-growing segment of Intercom’s $400 million ARR business.1 Fin is projected to account for half of Intercom’s total revenue in early 2027.
Intercom plans to expand Fin beyond customer service into sales and marketing, positioning it as a competitor to Salesforce’s Agentforce.1 The company’s success raises questions about the future of SaaS and the potential for domain-specific AI models to outperform generic, large language models.
The post Intercom Builds AI Model to Beat OpenAI & Anthropic in Customer Support appeared first on Archynewsy.