Kaltura conversational avatars vs Sume clips for a training portal
Kaltura launched real-time avatars with an SDK on 2026-03-12. For a training portal where lessons are fixed, Sume renders the avatar clips instead.
Kaltura announced real-time conversational avatars on 2026-03-12 as part of its AI Video Experience Cloud, with an SDK for embedding them in your own applications. Sume does not do live conversation; for a training portal whose lessons are written in advance, Sume Avatar 1.0 renders each lesson intro as a 4 to 60 second clip.
What Kaltura announced
The press release describes photorealistic, real-time conversational AI video agents that connect to Kaltura's Genie agents, video portals, virtual events and webinars, and virtual classrooms. It lists screen and camera comprehension, support for more than 30 languages, and a Conversational Agents SDK for custom workflows. Target uses include customer care, training, recruiting, financial onboarding and educational advising. Pricing is not disclosed.
That is a live-interaction product built into a video platform. If your learners need to ask a tutor questions, it is the right kind of tool.
Where rendered clips fit a training portal
Much of a course is not a conversation: a welcome, a module introduction, a reminder that a deadline moved. Those are the same for every learner, and they change on a schedule that suits a render pipeline.
| Need | Live conversational avatar | Sume avatar clip |
|---|---|---|
| Answer a learner's question | Yes | No |
| Module welcome, same for all | Overkill | Yes, 4 to 60 s |
| Approve wording before release | Hard, it improvises | Yes, script plus first-frame preview |
| Captions for muted viewing | Platform dependent | Inline captions, an add-on in the avatar estimate |
| Languages | 30+ listed | Script language of the avatar voice; check yours first |
Cost of a module intro
A 20-second intro at plus quality is 20 x $0.245 = $4.90. A 12-module course with one 20-second intro each is 12 x $4.90 = $58.80 at plus, or 12 x $3.68 = $44.16 at standard ($0.184 a second). Creating the avatar is a one-time $0.95. If you fix a typo in one module, you re-render that clip only.
Workflow
- Create one avatar handle for the course host.
- Write each intro as a script of at most 60 seconds.
- Create a preview, approve the first frame, then generate the final video.
- Turn on inline captions so the clip is readable without sound.
What to do
Split your portal's content into things learners watch and things they ask. Render the first list and leave the second to a live tool, if you need one at all.
Sources
Related posts
More in Comparisons
- Kling 3 two 15 s jobs make a 30 s clip: $4.30 silent, $6.40 with sound
Kling 4.0 allows 30 s in one clip. On Sume, Kling 3 stops at 15 s, so run two jobs and join them: $4.30 silent or $6.40 with audio, including the join.
- Kling 4.0's 15 references vs Sume: which video model takes the most
Kling 4.0 lists up to 15 references: 10 images and 5 videos. On Sume, Wan 3.0 takes 10 images, 5 videos and 5 audio clips; Omni 10 images and 3 short videos.
- Kling 4.0 Flash 3-20 s at 720p: the closest Sume models, priced
Kling 4.0 Flash is 3-20 s at 720p in early access. Which Sume models reach 3, 5, 10, 15 and 20 s at 720p, and what each clip costs, from Omni to Wan 3.0.
- Kling 4.0 Flash vs Gemini Omni Flash 1.1: two Flash models, two limits
Kling 4.0 Flash is 3-20 s at 720p in early access. Omni Flash 1.1 on Sume is 3-10 s up to 4K, 16:9 or 9:16, with a price per second. Limits side by side.
Written by Sume