Reka Rho-1: can you generate video with it today?
Reka Rho-1 is a 19B research preview announced October 5, 2026, with no public weights, API or demo link. For video you can run today, check Sume's catalog.
No. Reka Rho-1 is a research preview that Reka's page dates October 5, 2026, and the page lists contact details for partnerships but no public weights, no API and no demo link. You cannot call it from a script today, and Sume does not offer it. If the goal is a video clip this afternoon, use a model that is in a catalog you can query.
What is Rho-1?
Reka describes Rho-1 as a 19 billion parameter omni-reasoning model that understands, simulates and acts. It handles text, images, video and robot actions as tokens inside one neural network, rather than a pipeline of separate specialist models. Reka says it was trained from scratch on 320 H100 GPUs over three months.
In the demonstrations Reka describes, the model draws a static scene, locates objects in it, animates the scene into a video and edits the video, for example to change the weather, while explaining what it changed.
What are its limits, in Reka's words?
The research page is candid about where the preview falls short. These limits matter if you were hoping to replace a production video model:
- Long-horizon drift in extended rollouts.
- Weak temporal grounding across video frames.
- Brittle consistency when editing visuals.
- Native resolution capped at 672 by 384.
What can you run instead?
Sume's video router publishes a catalog with capabilities for each model, including resolutions, duration ranges and which inputs each accepts. Rho-1 is not in it, and the catalog is the only source for what a Sume job can use. The table lists a few rows from the Sume video router documentation.
| Sume model id | Resolutions | Duration | Input note |
|---|---|---|---|
| grok-imagine-video-1.5 | 480p, 720p | 4 to 15 s | Image-to-video only |
| seedance-2.5 | 480p, 720p, 1080p | 4 to 30 s | See catalog |
| wan-3.0 | 480p, 720p, 1080p | 2 to 30 s | Text, image or reference |
| gemini-omni-flash-1.1 | 360p, 720p, 1080p, 4K | 3 to 10 s | Text, image, reference or video edit |
Should you wait for Rho-1?
Treat the announcement as a signal about direction, not a tool. Models that generate video and also reason about what they generated are being built, but a 672 by 384 research preview without access is not a substitute for a clip you can hand to a client. Watch Reka's page for an API or weights, and keep your pipeline against a catalog you can query so a new model is a one-field change when it arrives.
Sources
Related posts
More in Models
- Grok Imagine video ids: three at xAI, one on Sume
xAI lists grok-imagine-video, 1.5 and 1.5-lite; Sume's catalog has only grok-imagine-video-1.5. A quick map so you send the id that exists.
- What is Grok Imagine Video 1.5 Lite, and what does Sume run?
Grok Imagine Video 1.5 Lite is xAI's low-price video model at $0.02 a second. Sume's catalog has grok-imagine-video-1.5 instead, image-to-video only.
- Whistle 16.9 MB on-device model vs Sume STT: which clips go where
Cactus Whistle transcribes 30-second clips on the device; Sume STT bills $0.01 per audio minute in the cloud. Which audio belongs on which, with a table.
- Whistle's 30-second window: passes for 5 minutes vs Sume STT
A 5-minute recording is ten 30-second Whistle passes on the device, or one Sume STT job at $0.05. How to plan chunking and when to skip it.
Written by Sume