Elorian opens Glimpse, a model that reasons by revisiting visual evidence
Elorian’s Glimpse research preview aims to reason visually by re-examining uncertain regions in images and video instead of describing the input only once.
AT A GLANCE
Key points
- Revisits images and video during reasoning
- Available through an API and interactive chat interface
- Input costs $0.60 and output $3 per million tokens
- Benchmark results are based on Elorian’s own evaluations
What is changing?
Elorian introduced its first model, Glimpse, as a research preview on October 16. According to the company, the model tracks objects through motion and viewpoint changes, understands how parts relate, and tries to predict physical consequences from observations. Glimpse is available through an API and an interactive chat interface.
Its “visual scratch space” can zoom into a region, rewatch a short video interval at a denser frame rate, and run code for measurement while reasoning. Elorian also says Glimpse can coordinate subagents that exchange text, visual artifacts, and internal representations.
In Elorian’s own evaluations, Glimpse beat larger models on selected scientific-chart, short-video-captioning, and open-vocabulary-counting tests, while trailing them on others. Input costs $0.60 per million tokens and output costs $3 per million. These are provider-reported results, not independent validation.
Explore the original sources
Source date: 16 October 2026.
Published by AI Radar: 17 Oct 2026, 13:15 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…