Guides
Runway's October Notes Read Like a Cost Manual: Project Credit Caps, Draft Tiers and Per-Second Prices
ยท RenderBob team
Runway's October release notes add models almost weekly. Read for cost, they also show the controls a studio needs once several models share one platform.
Runway's October release notes add models almost weekly. Read for cost, they also show the controls a studio needs once several models share one platform.
Where this comes from
We read a Releasebot summary of Runway's release notes. We could not open Runway's own help-centre page, so treat the details below as reported, not confirmed. The entries are dated from 1 to 21 October.
The cost controls
- Project credit caps (9 October, Enterprise). Workspace admins can set a credit budget per project, and generation pauses when the cap is reached. The account team has to enable it.
- Seedance 2.5 Draft Mode (2 October, all plans). It makes fast 480p drafts with post-generation enhancement, so iteration is quicker and cheaper. We covered the draft mode on 15 October.
- Grok Imagine Video 1.5 Lite (1 October, API). A faster, lower-cost option at 480p, 720p and 1080p for 1 to 15 seconds, without image or audio references.
- Kling models (13 October, API). Kling 2.5 Turbo Pro, Kling 3.0 and Kling O3, with pricing per second that varies by model.
- Image models, priced per image. Seedream 5.0 Flash is listed at 2 credits an image, and Nano Banana 2 Lite, added on 21 October, at 4 credits, with up to 14 reference images and prompts up to 5,500 characters.
The tooling side
The notes also list models added to Runway's MCP route for paid plans, including BiRefNet v2 background removal, Mirelo SFX 2.0 for sound-effect tracks, Fish Audio Drama 3 for voiceovers, Seedream 5.0 Flash and Ideogram 4.5. That is the same pattern as our 20 October MCP gateway checklist: an assistant can call several models through one door.
What the pricing does and does not tell you
Credit prices per image are easy to compare. Per-second video prices vary by model, and the notes we saw do not list them. The cheapest option per second can cost more per usable clip if it needs more attempts. That is our reading, not Runway's, and the only way to know is to count attempts on your own shots.
How to run a budget on a multi-model platform
- Set a cap per project, or a manual limit if the platform has none for your plan.
- Draft cheap, finish dear. Use a draft tier to settle the shot, then spend on the final.
- Log attempts per shot, not only total spend.
- Compare cost per usable clip, not cost per second.
- Check what each model leaves out. The lite video model above has no image or audio references.
What to watch
Runway's own notes with per-second prices, whether caps reach plans below Enterprise, and independent tests of usable-clip rates.
RenderBob 3 and this story
RenderBob 3 puts the same discipline in the edit: draft locally, route only approved shots to cloud models clip by clip, and keep each generated clip on its own lane so spend is visible per shot.
More from the blog
- Two Ten-Minute AI Video Claims in One Week: A Checklist for Judging Long-Form Generation
Higgsfield said on 13 October that its AI video now runs past ten minutes. VideoGen said on 20 October that Model 1 can assemble ten-minute videos. The methods may differ.
- Four Voice Releases in Three Weeks: Eleven v4, Grok Voice 3.0, Fish Audio Drama 3 and Voxtral 3 Compared
Four voice models have launched since 28 September: Eleven v4, Grok Voice 3.0, Fish Audio Drama 3, and Voxtral 3. Their headline numbers measure different steps.