Hi everyone!
Reka Edge is a clear bet on a very specific future: Physical AI needs frontier-level visual intelligence, but it also needs to be fast, lean, and deployable close to the real world.
The use of a 7B VLM, convolutional vision encoder, 3x fewer input tokens, robust video understanding, object grounding and tools, all aimed at edge deployments rather than massive cloud-only setups. This makes the whole thing more relevant to robotics, drones, wearables, automotive, and other systems where latency really matters.
You can try it out in the playground, run it through the API, self-host it with VLLM from Hugging Face, or call it through OpenRouter.
<a href