P-Video-Upscale Turbo: 65% Cheaper, $0.0035/s

The short answer
Pruna AI launched Turbo mode for P-Video-Upscale on November 17, 2026, claiming 65% lower cost than Standard (up to 30 fps) at prices starting at $0.0035 per second, with a slight quality trade-off. The model upscales video to 720p, 1080p, 2K or 4K and keeps audio and aspect ratio. It is available through Comfy partner nodes, Segmind and Runware.
On this page (10 sections)
What is P-Video-Upscale Turbo?
P-Video-Upscale Turbo is a faster mode of P-Video-Upscale, a video upscaling model from Pruna AI. Pruna announced it on November 17, 2026, saying it cuts cost by 65% compared with Standard (up to 30 fps), with a slight trade-off in quality.
In its announcement on X, Pruna says pricing starts at $0.0035 per second. The post links to a speed comparison, but it gives no numbers in the text.
The base model takes an existing clip and returns a higher-resolution H.264 MP4. It keeps the source aspect ratio and audio. For ad teams, the main appeal is cheap finishing of AI-generated clips that are often capped at 720p.
P-Video-Upscale Turbo release date and availability
Pruna announced Turbo on November 17, 2026, and the model is already reachable through several platforms. It runs as a partner node in ComfyUI, and it is listed on Segmind and Runware.
The Comfy documentation says the node, called Pruna P-Video Upscale, sits in the partner/video/Pruna category. It runs on Pruna's servers through the Comfy API, so no model download or local GPU is needed. You do need a Comfy API account with credits.
Segmind lists a playground and API for the model, and Runware lists it in its video upscaling collection. It is not announced as live in the AdGraam Playground.
P-Video-Upscale Turbo key features and specs
P-Video-Upscale outputs 720p, 1080p, 2K or 4K and preserves aspect ratio, audio and, by default, frame rate. Turbo is a toggle on the same node that trades some detail for speed. The Comfy docs describe the result as softer detail.
Four frames of increasing size representing 720p to 4K upscale targets
| Spec | Details |
|---|---|
| Targets | 720p, 1080p (default), 2K, 4K |
| Pixel area | About 0.9, 2.1, 3.7, 8.3 megapixels |
| Output frame rate | Source (capped at 60), or 16, 24, 25, 30, 48, 50, 60 |
| Input formats | mp4, mov, webm, up to 512 MiB |
| Output | H.264 MP4 with source audio |
| Frame caps | 2400 (720p), 1200 (1080p), 600 (2K), 300 (4K) |
| Approx. length at 24 fps | About 100, 50, 25, 12.5 seconds |
| Turbo | Faster, softer detail; Pruna says 65% cheaper than Standard up to 30 fps |
A few limits matter in practice:
- A source larger than the target comes back at its original size. A 1080p clip sent to 720p stays 1920×1080.
- Clips above 10 megapixels must target 2K or 4K.
- Outputs above 30 fps cost twice the per-second rate, including when
sourcekeeps a rate above 30. - Segmind notes that an upscaler cannot invent detail, so very blurry or heavily compressed clips gain less.
P-Video-Upscale Turbo pricing and free access
Pruna says Turbo starts at $0.0035 per second. Pricing differs by platform and is not uniform across the sources.
The Comfy docs say billing is per second of output video and depends on the target, and they point to a partner node pricing page for current rates. They also describe Turbo as "at the same price" in the node controls table, which does not match Pruna's 65% claim. Check the current Comfy rate before budgeting.
Runware lists P-Video-Upscale at $0.01 per second for 720p, up to 30 fps. For comparison, Runware lists these upscalers:
| Model | Price | Notes |
|---|---|---|
| P-Video-Upscale (Runware) | $0.01/second | 720p, up to 30 fps |
| ByteDance Video Upscaler | $0.00265/second | 720p, under 30 fps |
| BytePlus Video Enhancement Standard | $0.003434/second | 240p to 720p |
| Topaz Labs Starlight Precise 2.5 | $0.08/second | 1K |
| FLUX 3 Upscaler | $0.15/second | Precise, 1080p |
No free tier is mentioned. At $0.0035 per second, a 15 second clip would cost about $0.05 at Pruna's quoted starting rate, though the rate for higher targets is not stated.
How to use P-Video-Upscale Turbo for ad creatives
Turbo suits work where you need many clips cheaply and can accept softer detail. Think draft passes, internal reviews and test variants, with Standard reserved for the clips that win.
A desk with a node-based video workflow on a monitor
A practical split for a D2C team:
- Hooks and test variants: upscale many 720p AI clips with Turbo to check how they read on a phone before spending on finals.
- Winning ads: rerun the best performers on Standard at 1080p or 2K.
- Social exports: clean up compressed clips before reposting, as Segmind suggests.
- Chained workflows: the Comfy node accepts video from a text-to-video or image-to-video node, so upscaling can sit at the end of a generation graph.
Keep clips short. At 4K the cap is 300 frames, about 12.5 seconds at 24 fps, which covers most ad cuts but not long edits. Keep fps at source unless you want interpolation, since anything above 30 fps doubles the rate.
For teams that chain image, video and upscale models, AdGraam Flow Builder lets you link those steps into a workflow you can rerun, and the AdGraam Playground lets you run several models on the same input and compare the results. For hook testing, see our guide to hook rate and hold rate.
What to watch next
Pruna's post links to a speed chart but gives no numbers in the text, and "slight" quality loss is Pruna's own word. Watch for side by side samples of Turbo and Standard before relying on Turbo for finals.
Also watch the pricing mismatch. Pruna says Turbo is 65% cheaper, while the Comfy docs call it the same price. Whichever platform you use, confirm the billed rate on a short test clip first.
Pruna has other models listed elsewhere. A Wiro listing shows P-Image Try-On at $0.015 for the first garment image in Standard and $0.008 per garment image in Turbo. A post in a Stable Diffusion Facebook group says P-H3 Image, which it describes as built on MiniMax H3, generates images up to 4K and edits photos from a one sentence instruction. Pruna has not confirmed that in anything covered here.
What people are posting
P-Video-Upscale just got faster and cheaper! ⚡ Introducing Turbo mode: faster video upscaling at 65% lower cost compared to Standard (up to 30 fps), with a slight trade-off in quality. Here's what you need to know: - 𝗣𝗿𝗶𝗰𝗶𝗻𝗴: Starting at just $0.0035/s - 𝗦𝗽𝗲𝗲𝗱: https://t.co/KTVaonMVyF
Key takeaways
- Pruna announced Turbo mode for P-Video-Upscale on 2026-11-17, with pricing starting at $0.0035 per second.
- Pruna says Turbo costs 65% less than Standard for output up to 30 fps, with a slight quality trade-off.
- Targets are 720p, 1080p (default), 2K and 4K, with source aspect ratio and audio preserved.
- Frame rate can be kept or interpolated up to 60 fps, but output above 30 fps costs twice the per-second rate, per Comfy docs.
- Input accepts mp4, mov or webm up to 512 MiB, with per-target frame caps from 2400 frames at 720p down to 300 at 4K.
- Runware lists the model at $0.01 per second for 720p up to 30 fps.
Compare 200+ AI video and image models side by side
Run Kling, Veo, Sora, Seedance, Flux, Ideogram and Recraft in one Playground, or turn any product URL into ready-to-launch ads.
Frequently asked
What is P-Video-Upscale Turbo?
It is a faster mode of Pruna AI's P-Video-Upscale video upscaler. Pruna says it lowers cost by 65% compared with Standard for output up to 30 fps, with a slight quality trade-off.
How much does P-Video-Upscale Turbo cost?
Pruna says pricing starts at $0.0035 per second. Runware lists the model at $0.01 per second for 720p up to 30 fps, and Comfy bills per second of output with rates that depend on the target.
What resolutions does P-Video-Upscale support?
It upscales to 720p, 1080p, 2K or 4K, which correspond to about 0.9, 2.1, 3.7 and 8.3 megapixels. It keeps the source aspect ratio and audio.
How do I access P-Video-Upscale?
It runs as a Pruna P-Video Upscale partner node in ComfyUI through the Comfy API, which needs an account with credits. It is also listed on Segmind and Runware. It is not announced as live in the AdGraam Playground.
Does Turbo lower quality?
Pruna describes a slight quality trade-off, and the Comfy docs describe softer detail. Segmind suggests Turbo for bulk or draft passes and Turbo off for clean finals.
Sources
- Pruna P-Video Upscale: Upscale Video to 720p, 1080p, 2K ...docs.comfy.org
- Pruna P Video Upscale APIsegmind.com
- Best Video Upscaling AI Modelsrunware.ai
- P Image Try On API: Pricing, Sample Image, Docswiro.ai
- Pruna P-H3 Image in ComfyUI | MiniMax H3 for 4K ...facebook.com
More AI model news & creative guides
HeyGen Avatar 6 Is Live: Native 2K, #1 on Arena
HeyGen Avatar 6 is live with native 2K frames and an 84.4% indistinguishable score. Get the spec table, benchmark breakdown, ad workflow tips and access details.

Seedance MCP at $0.0918/Sec: YouArt's Pricing Claim
YouArt says its Seedance MCP costs $0.0918/sec pay as you go. Get the price claim vs BytePlus list rates, a spec table, ad workflow notes and access details.

Muse Code 3D: Native GLB Generation, Limits & Access
Meta added native text and image to GLB 3D generation to Muse Code on Nov 20, 2026. Get the spec table, ad workflow breakdown, plan limits and access details.
Best AI Ad Generators for D2C Brands in 2026
Ten AI ad generators compared for D2C and e-commerce brands: what each makes, real starting prices, free options, and which suits Indian brands paying in rupees.
Break Even ROAS for COD Orders: How RTO Changes the Math
How to work out break even ROAS when a share of COD orders comes back as RTO, with worked rupee examples, a prepaid and COD mix, and why Ads Manager ROAS overstates COD.
What Indian D2C Brands Run on Meta: 2,790 Ads Analyzed
We studied 2,790 Facebook and Instagram ads from 14 Indian brands and their 747 longest-running ads: formats, angles, offers, languages and CTAs that last.
Your move
See your product as a winning ad
Turn a product URL into static ads and UGC-style videos, rebuild the competitor ads that already work, and read your Meta data to know what to make next.
AdGraam News DeskAI video and image news
The AdGraam News Desk (part of AdGraam, also searched as Adgram or Adgram AI) tracks AI image and video model launches, leaks and rumors from labs, leakers and creators, and checks each claim against its sources. Unconfirmed stories say so, and each page is updated as its story develops.