Elorian launches Glimpse visual reasoning model

Elorian released its first AI model, Glimpse, as a research preview with API access and a chat interface, the company said in a blog post, as TechCrunch first reported. Elorian says the visual reasoning model matches or beats GPT-6 Astra and Claude Opus 5.5 on several benchmarks it ran itself, at a much lower price.

Glimpse can revisit visual inputs while reasoning. It can crop an image, inspect selected video frames and run code to measure or compare visual information. Elorian calls this process a visual scratch space. The model can also coordinate subagents that exchange text and visual material.

In Elorian's own tests, Glimpse exceeded GPT-6 Astra on four evaluations covering scientific charts, open vocabulary counting, short form video captions and manipulation events. Input costs $0.60 per 1 million tokens, while output costs $3 per 1 million tokens. The company says a small team trained the model using a few hundred GPUs and an open weights base model.