Elorian launches Glimpse, an AI model that keeps looking back at images while it reasons

Elorian, a Palo Alto AI startup run by former Google DeepMind researcher Andrew Dai, released its first model on October 16, 2026, TechCrunch reported. Called Glimpse, it is available as a research preview through an API and a chat interface.

Most multimodal models turn an image into tokens once and then reason about it in text. Glimpse can go back to the picture or video partway through its reasoning, cropping into a region, pulling particular video frames, or running code to measure and compare what it sees. Elorian calls this a visual scratch space. Dai said the idea came partly from asking a top model who was winning a board game from a photo, which it could not answer.

Elorian says a small team trained Glimpse on a few hundred GPUs and that it is much smaller than the frontier models it is compared with. In the company's own tests it beats GPT-6 Astra on scientific chart reasoning and short video captioning. Pricing is $0.60 per million input tokens and $3 per million output tokens. The startup raised a $55 million seed round at a $300 million valuation before launching, and is first targeting robotics firms, science labs and media companies with large amounts of video.

Why it matters

Models that can actually inspect visual details could make AI far more useful for charts, diagrams, video review and, eventually, robots.

Source: TechCrunch