Elorian launches Glimpse, its first AI model for visual reasoning
Palo Alto AI startup Elorian has released Glimpse as a research preview, priced at $0.60 per million input tokens.
Stock photo for illustration only, not from the actual event
- Elorian has released its first AI model, Glimpse, as a research preview.
- Pricing is set at $0.60 per million input tokens and $3 per million output tokens.
- Glimpse outperforms GPT-6 Astra in scientific chart reasoning and short video captioning.
- The company's long-term goal is to transition from visual perception to direct robotic control.
Palo Alto-based artificial intelligence startup Elorian, which previously raised a 55 million dollar seed round at a 300 million dollar valuation before launching any product, has released its first model on Friday, according to an exclusive report by TechCrunch.
The newly unveiled model is named Glimpse, launching as a research preview with API access while its chat interface remains waitlist-only. For now, the rollout consists of a results table and a core thesis: that today's AI models, despite their proficiency with language and code, still cannot genuinely see or comprehend images.
For former Google DeepMind researcher and Elorian CEO Andrew Dai, one of the most definitive signs that artificial intelligence still lacked true visual comprehension occurred during a game night.
Regarding its pricing structure, Glimpse costs 0.60 dollars per million input tokens and 3 dollars per million output tokens, making it cheaper than mid-level models such as Sonnet and Sol, yet more expensive than lightweight alternatives like Haiku and Luna.
Stock photo for illustration only, not from the actual event
Elorian's entry into deep visual reasoning represents a crucial challenge to the current AI landscape. While conventional models excel at surface-level image descriptions, advanced logical reasoning and spatial intelligence remain significant hurdles for automation and robotics. Targeting industries rich in video assets, such as scientific labs and robotics firms, marks a strategic commercial positioning.
Based on evaluations conducted by Elorian, Glimpse outperforms even GPT-6 Astra when handling scientific chart reasoning and short-form video captioning, alongside demonstrating competitive performance across other visual reasoning benchmarks.
Elorian initially emerged from stealth six months ago through a funding round co-led by Striker Venture Partners, Menlo Ventures, and Altimeter Capital, with additional participation from Nvidia and Google chief scientist Jeff Dean.
The startup's initial target market consists of businesses sitting on vast amounts of video data, including robotics enterprises, scientific laboratories, media outlets, and creative agencies. Over the longer term, Elorian aims to evolve from mere perception to active execution by constructing environmental scene models to determine actions, including the direct control of robots. Next year, Dai noted, the company intends to build what it considers the world's leading visual reasoning model, targeting top performance across every visual reasoning metric.
Source: TechCrunch
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment