Multimodal AI for enterprise data and on-device visual intelligence
HQ: Sunnyvale, California, United States; distributed team | Founded: 2022 | Employees: Not disclosed in the reviewed sources | Stage: $110M investment announced July 2025; formal series and exact valuation not confirmed in company release | Website: https://reka.ai/
Reka develops multimodal models and applications for text, images, video, and audio. Its July 2025 release identifies Sunnyvale headquarters and describes Reka Vision for visual-content search and reasoning, Reka Research for web/private-document synthesis, and Reka Flash as a multimodal model.
Radical Ventures lists five founders: Dani Yogatama, Cyprien de Masson d'Autume, Qi Liu, Yi Tay, and Mikel Artetxe; its page identifies Yogatama as CEO and de Masson d'Autume as CTO. Dealroom records 2022 as the founding year.
The June 2023 company announcement disclosed $60M from DST Global Partners, founding investor Radical Ventures, and others. July 2025 brought a $110M investment with NVIDIA and Snowflake among backers, without an exact valuation or lead role in the company release. Current research pages describe open-weight Reka Edge and EdgeQ for on-device visual AI. Enterprise platforms and local research implementations are separate offerings with potentially different availability and economics.
Reka is a multimodal research-and-product business rather than another text-only assistant. Enterprise platforms address visual content and information synthesis, while 2026 research pushes visual intelligence toward local device execution. That provides a clearer identity than competing only on general chatbot benchmarks.
The current signal is EdgeQ, highlighted September 22, 2026: Reka Edge implemented for Qualcomm's Hexagon NPU. Official materials describe an open-weight 7B model, local deployment, and APIs. Latency, thermal, and leadership comparisons remain company measurements, not independently confirmed results in this research.
Company announcements establish $60M in June 2023 and $110M in July 2025, a $170M subtotal. Some trackers report $168M; this profile uses explicit primary announcements without pretending to reconcile the cap table. Editorial assessment: enterprise visual data and edge deployment are useful wedges, but major model platforms, open alternatives, and private-data integration remain substantial competition.
Category: AI & Machine Learning Startups
Browse: Home | Blog | About | All Startups A-Z | AI & Machine Learning Startups | View full profile on StartupWiki