Andrew Dai's first model for the startup revisits images and video as it reasons, with company-reported benchmark wins and clear gaps against frontier systems.
By [RuntimeWire Staff](https://runtimewire.com/author/runtimewire-staff)
· Published
Primary source: [TechCrunch](https://techcrunch.com/2026/10/16/elorians-first-ai-model-wants-to-reason-with-images-not-just-about-them/)
Why it matters #
Elorian is testing whether a smaller model built around repeated visual inspection can earn a place in robotics, science and media workflows. Its benchmark wins are narrow, company-reported results, and its own table shows important losses to GPT-6 Astra.
Elorian released its first model, Glimpse, on October 16th, built around a simple idea from co-founder and CEO Andrew Dai: when a model is unsure what an image shows, it should inspect the image again rather than keep reasoning from its first impression. The model is available as a research preview through an API; Elorian's chat interface is waitlist-only, according to TechCrunch.…