Glimpse by Elorian - A visual thinking model that reasons about the real world
by•
Glimpse is Elorian’s visual thinking model, a research preview that reasons about the physical world from what it sees. It follows objects through motion and changing viewpoints, and uses tools to write code, transform images, and watch videos. Elorian says it beats GPT-6 Astra on four multimodal benchmarks. API at $0.60 per million input tokens and $3.00 per million output. Free to try on the Elorian platform.
Replies
Hey Product Hunt 👋
Most vision models describe what’s in an image or video, but they struggle to reason about what’s physically happening: tracking one object through motion, counting across a shaky video, or predicting what happens next.
Solution
Glimpse is a visual thinking model from Elorian that reasons about the physical world from what it sees. Per Elorian, it zooms into regions, rewatches short clips at a higher frame rate, and tracks features over time. The platform shows this “visual scratch space” as it works, so you can see the evidence behind an answer.
Key features
Visual reasoning that revisits and refines what it sees as it thinks
Agent behavior: tools that write code, transform images and watch videos
Subagents that share text and visual artifacts
Demos include tracking a dolphin in a pod, counting puppies in a handheld video, and robotics data labeling and planning
API and chat UI, with a free try on the platform
Pricing per Elorian: $0.60 per million input tokens and $3.00 per million output tokens
Links
Try it on the Elorian platform
Models page
Launch blog