Glimpse model claims frontier multimodal performance at 17x lower token prices
dustinvtran · x · 2026-10-17
- The developer claims Glimpse reaches frontier-level multimodal performance, matching or exceeding GPT-6 Astra on 5 of 12 benchmarks at 17x lower token prices
- Smaller and cheaper than Gemini 3.8 Flash while dominating it on real-world vision workloads (WGO Bench 42.70 vs 33.99, FSC-147 counting 83.36 vs 67.06)
- Pricing: $0.60/1M input tokens, $3.00/1M output tokens
- Enabled by a new visual reasoning technique freeing the model from text-only chain-of-thought
The attached table benchmarks Glimpse against GPT-6 Astra, GPT-6.1 Sol, Opus 5.5 and Gemini 3.8 Flash across robotics, visual perception, scientific reasoning, video captioning, temporal reasoning and counting. These are self-reported figures pending third-party verification.
More from Models
- Grok Bots Roll Out on Grok Web and Mobile, Filling the Gap OpenAI's Dots Leaves Open — testingcatalog · 2026-10-17
- Meta terms reportedly grant itself rights to scrape the open web for AI training — VeryWellVersed · 2026-10-17
- Codex users lose remote usage tracking, leaving only an edit button — No-Butterscotch-218 · 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