Elorian Launches Glimpse, a 'Visual Thinking' Model Claiming Frontier-Level Benchmarks

AndrewDai · x · 2026-10-17

Startup Elorian launched Glimpse, pitched as a visual thinking model that tracks identity through motion, reasons across viewpoints, and predicts physical consequences. The team claims frontier-level multimodal benchmark performance vs GPT-6 Astra and Claude Opus 5.5 with a smaller, cheaper model, plus end-to-end multimodal tool use and coordinated subagents. Free on its chat interface with an API; claims are unverified.

Related event: Elorian Launches Visual Thinking Model Glimpse, Claiming Astra-Level Performance on a Few Hundred GPUs(19 posts)→

Original post →

More from Models

• Ex-Gemini and Apple MM-1 builders launch Glimpse, a visual-reasoning-only model — heyshrutimishra · 2026-10-17
• Meta terms reportedly grant itself rights to scrape the open web for AI training — VeryWellVersed · 2026-10-17
• Dev asks why non-tech users prefer ChatGPT when Claude is 'far better' — prasenx · 2026-10-17
• ChatGPT's annoying regressions: garbled input, broken inline images, constant timeouts — ChristianKl · 2026-10-17
• Model tweaks barely move MMLU but easily break generative tasks — aryaman2020 · 2026-10-17
• Claude reverse-engineers 1977 DES test vectors to a PDP-11 rand() seed — chrisrohlf · 2026-10-17

Models channel →