The addition of vision allows the model to perform optical character recognition without prompt engineering and can improve performance for text-only tasks, according to a technical review. The model's architecture is designed to remain computationally efficient and inexpensive even when processing complex visual inputs for yes/no, choice, or score questions.
Wednesday, October 21, 2026
Story
Liquid AI d1 Matches GPT-6.1 Sol on 4 of 6 Tasks at 200x Lower Cost
AI15 days ago
The company's first decision model now supports images, text, or both as inputs to automate industrial workloads such as inspecting circuit boards. Liquid AI tested d1 against OpenAI's GPT-6.1 Sol and Anthropic's Claude Opus 5.5 across six real applications, including filtering support tickets, where d1 matched or beat GPT-6.1 Sol in four instances. The model reduces costs by 19x to 200x compared to its competitors and produces text decisions in 200 to 300 ms using a single forward pass without generating tokens.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.