Katanemo has introduced Arch-Function, a collection of open-source large language models (LLMs) designed for ultra-fast function-calling tasks essential for agentic applications in enterprises. According to reporting from VentureBeat, these models operate nearly 12 times faster than OpenAI’s GPT-4 and significantly outperform offerings from other competitors, while also providing substantial cost savings.
Arch-Function builds on Katanemo’s previous release, Arch, which serves as an intelligent prompt gateway, enabling developers to create secure and efficient generative AI applications. The new LLMs, based on Qwen 2.5, can understand complex prompts and perform digital tasks by interacting with external systems. They are particularly adept at executing functions through API calls, allowing for tailored workflows across various domains, from insurance claims to marketing campaigns.
The models demonstrate significant throughput improvements and cost efficiencies, achieving a 44x cost reduction compared to GPT-4. While full benchmarks are pending, initial results indicate that these LLMs are a strong choice for real-time applications, aligning with the projected growth of the AI agents market, expected to reach $47 billion by 2030.