Where is Reka AI headquartered?

Reka AI is headquartered in San Francisco, USA. The company has a team of 51-200 people.

When was Reka AI founded?

Reka AI was founded in 2022. It is based in San Francisco, USA.

Who is the CEO of Reka AI?

Dani Yogatama is the CEO of Reka AI. The company was founded in 2022.

Where is Reka AI located?

Reka AI is headquartered in San Francisco, USA.

What funding stage is Reka AI at?

Reka AI is currently at the Growth Stage stage.

How many employees does Reka AI have?

Reka AI has approximately 51-200 employees.

Reka AI

Category: Foundation Models / LLMs

Reka is an AI research lab building foundational intelligence for the physical world—natively multimodal omni models and infrastructure for robotics, wearables, video understanding, and real-time physical reasoning. Reka AI was founded in 2022. The company is led by Dani Yogatama. Based in San Francisco, USA. Team size: 51-200. Total funding raised: $168M. Latest round: Series B ($110M, backed by NVIDIA and Snowflake). Key investors include Snowflake Ventures, NVIDIA, DST Global Partners, Radical Ventures, Nat Friedman.

Founded: 2022
Headquarters: San Francisco, USA
Team size: 51-200
Total funding: $168M

Value proposition

Reka offers powerful, efficient, and customizable multimodal AI models that enable businesses to build next-generation AI applications. Their key value lies in providing frontier-level performance with cost-effective, flexible deployment options that can handle diverse data types, bridging the gap between AI potential and real-world application.

Products and solutions

Reka Labs: R&D of foundational intelligence for the physical world, including World Language Action Model (WLAM)., Reka Infer: Inference engine and API for multimodal AI (tryinfer.com)., Reka Vision: Video processing infrastructure to tag, reason, search, and clip large video volumes (API/MCP/app)., Reka Claru: Training-data platform for physical AI—egocentric video, robotics trajectories, world-model footage., Reka Edge / Edge 2: 7B vision-language model for on-device/edge physical AI (open weights on HF)., Reka Flash / Core: Multimodal model family (API-priced); Reka Research agentic web/doc research; Reka Speech.

Unique value

Reka's core innovation lies in its 'multimodality-native' approach, building models from the ground up to inherently understand and reason across a combination of text, images, video, and audio. Unlike competitors who may add modalities to existing language models, Reka focuses on creating a unified understanding. They also provide a spectrum of model sizes (Core, Flash, Edge) optimized for different use cases, from powerful cloud-based reasoning to efficient on-device performance.

Target customer

The primary target customers are enterprises seeking to integrate advanced AI capabilities into their workflows. They also cater to developers, creators, and organizations that require customizable and deployable AI models, offering cloud, on-premise, and on-device solutions.

Industries served

Enterprise Software, Cloud Computing, Media & Entertainment, E-commerce, Manufacturing, Gaming Technology

Technology advantage

Reka's key advantage stems from its founding team of distinguished scientists from top research labs like Google DeepMind, Meta AI, and Baidu, providing deep expertise in AI model development. The company operates a full-stack strategy, training its own proprietary foundation models and building applications on top of them. This allows for greater control, customization, and the development of specialized 'agentic AI' capable of performing complex, multi-step tasks.

How they differentiate

Reka AI differentiates itself by focusing on developing highly efficient, multimodal AI models (capable of processing text, images, video, and audio) that are customizable for enterprise applications. They offer a suite of models of varying sizes (Core, Flash, Edge) to cater to different performance and cost needs, from powerful cloud-based models to lightweight ones that can run on-device. Their proprietary 'distillation' technology allows them to create smaller, specialized models for specific use cases, which is a key distinction from competitors who often focus on large, general-purpose models.

Main competitors

OpenAI, Anthropic, Cohere

Key partnerships

Moonvalley (Jun 2026): Merged team into Reka to advance physical AI / WLAM; Mateusz Malinowski and Mikołaj Bińkowski joined research leadership., Snowflake: Multimodal models in Snowflake Cortex; also Series B investor., Oracle: OCI training + Oracle Cloud Marketplace; Oracle Defense Ecosystem (Oct 2025)., NVIDIA: Series B investor; Vision models with NVIDIA AI blueprints., Shutterstock: Intelligent metadata for image/video library., Alibaba Cloud: Models on Alibaba Cloud marketplace., Carahsoft: U.S. Public Sector reseller., OpenRouter: Reka Edge available via OpenRouter (Apr 2026).

Notable customers

Snowflake (also an investor and partner, offering Reka's models to its customers), NVIDIA (partner for integrating Reka's vision models with NVIDIA's AI blueprints), Telkom Indonesia (strategic partner), Alibaba Cloud (strategic partner)

Major milestones

Founded in 2022 by leading AI researchers from Google, Meta, and DeepMind., Emerged from stealth in June 2023 with a $58 million Series A., Launched multimodal suite (Core/Flash/Edge); Reka Vision and Reka Research GA around Series B., Jul 2025: $110M Series B (NVIDIA, Snowflake) at >$1B unicorn valuation., Jun 9 2026: Merged with Moonvalley, adding Veo-contributor research leadership and pivoting GTM to physical AI (Labs/Infer/Vision/Claru)., Ongoing open releases (e.g., Reka Edge weights, CS2-10k, Research-Eval) and physical-AI research through Aug 2026.

Growth metrics

While specific revenue figures are not public, Reka AI achieved a unicorn valuation of over $1 billion in its latest funding round. The company has rapidly developed and released a suite of multimodal models (Core, Flash, Edge, Spark) that are competitive with leading models from major AI labs.

Market positioning

Reka is positioned as a research-driven physical-AI lab: natively multimodal omni models plus full-stack pillars (Labs, Infer, Vision, Claru) for robotics, wearables, video intelligence, and real-world agents. Unicorn (~$1B+, Jul 2025) competing with frontier labs while partnering with NVIDIA, Snowflake, and Oracle; post-Moonvalley merger focused on World Language Action Models and edge-deployable VLMs.

Geographic focus

Reka AI is headquartered in Sunnyvale, California, and operates with a globally distributed team. While its primary focus is the US market, it has also established strategic partnerships in Asia with companies like Telkom Indonesia and Alibaba Cloud, indicating a strategy for global expansion.

Patents and IP

No public patents have been found. The company's intellectual property appears to be concentrated in its proprietary model architectures, training methodologies, and the collective expertise of its research and engineering teams.

About Dani Yogatama

Dani Yogatama is a distinguished figure in the AI industry, currently serving as the Co-founder and CEO of Reka AI. His career is marked by significant contributions to AI research and development, with a focus on natural language processing and deep learning. Before establishing Reka AI, he was a Senior Staff Research Scientist at DeepMind, where he made substantial advancements in large-scale language models. His experience also includes a tenure at Baidu's Silicon Valley AI Lab, further solidifying his expertise in cutting-edge AI technologies. Dani holds a Ph.D. in Language and Information Technologies from Carnegie Mellon University, a testament to his strong academic foundation. His leadership at Reka AI is driven by a vision to push the boundaries of AI and create impactful, real-world applications.

More Foundation Models / LLMs companies

Official website: https://reka.ai