Palo Alto — Elorian Launches Glimpse AI Model for Visual Reasoning

Elorian released its first AI model, Glimpse, on Friday as a research preview with API access and a chat interface, according to TechCrunch. The Palo Alto startup says the system can crop into an image, pull specific video frames, or run code to measure what it is looking at while it reasons — a feature Elorian calls a "visual scratch space."

How does Glimpse differ from other multimodal models?

Most multimodal systems convert an image to tokens once and then reason about it entirely in text, TechCrunch reported, comparing the method to describing a photo from memory. Glimpse instead returns to the source image or video mid-reasoning, similar to a person pinching to zoom on a phone photo to double-check a detail. OpenAI added a comparable crop-and-zoom step to its o-series models in 2025, per TechCrunch, but Elorian says it built its entire architecture around the mechanism.

What prompted the design?

Elorian CEO Andrew Dai, a former Google DeepMind researcher, told TechCrunch the idea crystallized during a board-game night last year, when he photographed a finished game and asked a frontier model who had won. It could not answer.

"To an adult, it's quite easy to understand what's going on with minimal text, but these models aren't able to intuit what the symbols mean," Dai said, per TechCrunch. "Humans can look at Ikea diagrams, with no text, and assemble Ikea furniture, right? Why can't AI models?"

How does Glimpse perform against GPT-6 Astra?

According to Elorian's own evaluations, cited by TechCrunch, Glimpse outperforms GPT-6 Astra on scientific chart reasoning and short-form video captioning, with competitive results on other visual reasoning tasks. Elorian says a small team trained the model on a few hundred GPUs and that it is substantially smaller than the frontier systems it is benchmarked against. Those figures come from Elorian, not an independent third-party test, and TechCrunch's report does not include outside verification.

What does Glimpse cost, and who is it built for?

Elorian is pricing Glimpse at $0.60 per million input tokens and $3 per million output tokens, TechCrunch reported — cheaper than mid-tier models such as Sonnet and Sol but pricier than lightweight models like Haiku and Luna. The company's initial targets are robotics firms, scientific labs, media companies, and creative studios that work with large volumes of video, according to the report.

What is Elorian's timeline so far?

Elorian came out of stealth six months before Friday's launch, with a seed round co-led by Striker Venture Partners, Menlo Ventures, and Altimeter Capital, and participation from Nvidia and Google chief scientist Jeff Dean, TechCrunch reported. That round valued the company at $300 million before it had shipped a product.

What to watch

• Whether independent benchmarks confirm Elorian's chart- and video-reasoning claims against GPT-6 Astra.
• Adoption among robotics and scientific-lab customers Elorian is courting first.
• Elorian's stated 2027 goal, described by Dai to TechCrunch as building toward models that choose actions from a scene and directly control robots.
• Pricing moves from competitors now that Glimpse sits between lightweight and mid-tier model costs.