Correction, 2026-10-01: an earlier version of this page put a Wan 2.5 clip at about $0.65 at pack rate (102 credits is about $1.02), suggested upscaling a Wan clip with Topaz (8frame's upscalers work on images, not video), and gave VRAM figures we could not source. The local hardware section now quotes the Wan 2.2 model card, and the page notes that the Wan tool on 8frame now starts on Wan 3.0, which costs less than Wan 2.5 at 1080p. It also counted Wan 2.2 clips on 8frame without saying that Wan 2.2 there needs a start image or an input video and does not take text alone; it now says so.
Wan is the AI video family from Alibaba, and its open branch stops at Wan 2.2, which you can run on your own hardware for free, in theory. Wan 2.5 is API-only, so browser or nothing. This guide is about the practice: what running Wan locally actually demands in GPU VRAM, and how to skip all of it by using Wan 2.5 online on the 8frame canvas. By the end you'll know which path fits your hardware and your patience.
What you'll end up with is 1080p AI video without owning a workstation. On 8frame that's watermark-free output at 102 credits a 5-second clip, about 9 clips a month on the $19 Starter plan, with no download, no ComfyUI graph to debug, and no GPU to buy. Wan 3.0, which the Wan tool now starts on, is cheaper: 95 credits at 1080p and 24 at 480p.
What Wan is
Wan is Alibaba's video generation model family, and Wan 2.2 is the last release with published model files, so anyone with sufficient hardware can download and run that version locally. Wan 2.5 and Wan 3.0 live behind APIs like Veo or Kling do. That earlier openness is why Wan anchors the budget end of the market. For the full background, see What is Wan.
The local route: what "free" actually costs in VRAM
Running Wan locally is genuinely free of per-clip charges. It is not free of hardware. Here is what Alibaba's own Wan 2.2 model card says, because "no GPU required" is only meaningful once you know what the GPU requirement would otherwise be.
- Wan 2.2 ships three models: T2V-A14B (text to video), I2V-A14B (image to video) and TI2V-5B (both, in one smaller model), under the Apache 2.0 licence.
- TI2V-5B makes 720p at 24 frames per second. Its documented command "can run on a GPU with at least 24GB VRAM (e.g, RTX 4090 GPU)", and it "can generate a 5-second 720P video in under 9 minutes on a single consumer-grade GPU".
- The 14B models run with model offloading in the documented commands; the card says the offloading options can be dropped only "on a GPU with at least 80GB VRAM".
Community builds squeeze these models onto smaller cards with quantization and offloading; we have not tested them, so we quote only the official figures. In practice, the documented local route means a 24 GB consumer GPU at minimum, a ComfyUI or command-line install, model downloads measured in tens of gigabytes, and minutes per clip. None of that is a criticism of the model. It's just the entry ticket, and most marketers and creators don't have a 24 GB card sitting idle.
The online route: Wan 2.5 on the 8frame canvas
The alternative is to let someone else own the GPU. On 8frame, Wan 2.5 runs hosted, in the browser, on every plan. No install, no VRAM math.
-
Open the canvas and add a Wan node, then pick Wan 2.5. The Wan tool starts on Wan 3.0, so switch the model if you want 2.5 specifically. No download, no environment setup. The model runs server-side.
-
Write a prompt or drop a reference frame. Wan handles both text-to-video and image-to-video. Keep prompts concrete: subject, motion, camera, lighting. Example: "slow push-in on a coffee cup steaming on a wooden table, morning light, shallow depth of field." For tested structures aimed at cheap b-roll, see Wan 2.5 prompts for free b-roll.
-
Generate. Output is watermark-free at 1080p, 5-second clips, at 102 credits each. The $19 Starter plan's 1,000 monthly credits cover about 9 Wan 2.5 clips or 20 Wan 2.2 clips, enough for prototyping, mood tests, and student projects. Wan 2.2 on 8frame starts from an image or a video, not from text alone; for a text prompt, use Wan 2.5 or Wan 3.0.
-
Upgrade only the clips that earn it. When a Wan draft is worth finishing, use the winning frame as the start image for another model such as Seedance or Kling, from the same credit pool. Topaz, Clarity and BRIA can upscale that frame as a still; 8frame has no video upscaler, so pick the delivery resolution before the final run.
The reason this beats a local install for most people is not just the missing GPU. It's that on the canvas Wan sits next to every other model, so the moment a Wan draft proves a shot works, the stronger model that finishes it is one wire away. A local ComfyUI setup gives you unlimited free Wan generations, but only Wan, and only after you've paid for the hardware and learned the graph.
Local vs online: which to pick
| Your situation | Best route |
|---|---|
| You own a 24 GB+ GPU and like tinkering | Local (ComfyUI, open weights) |
| You want unlimited free generations and control | Local, if you have the hardware |
| You have no capable GPU | 8frame online, from $19/month |
| You want to prototype without any install | 8frame online |
| You need the winning clip finished at higher quality | Route to a stronger model on the 8frame canvas |
| You want Wan next to Veo, Kling, Seedance for comparison | 8frame canvas |
Cost math
Say you need 8 quick b-roll drafts this month to test hooks before committing to a real production model.
- Local: $0 per clip, but only after a 24 GB GPU (hundreds to well over a thousand dollars if you're buying), a ComfyUI setup, and the time to tune offloading. Free per clip, expensive per entry.
- 8frame hosted, Wan 2.5: 102 credits per clip, so 8 drafts cost 816 credits out of the 1,000 that come with the $19 Starter plan. No hardware, no setup.
- 8frame hosted, Wan 3.0 at 480p: 24 credits per 5-second clip, so the same 8 drafts cost 192.
- Credit packs, if you need overflow: top-ups start at $10 for 1,000 credits, which puts a Wan 2.5 clip at about $1.02 at pack rate and a 480p Wan 3.0 draft at about $0.24.
If you already have the workstation and enjoy owning the pipeline, local Wan is a legitimately free and powerful option. If you don't, $19 of hosted credits gets you the same model without buying a GPU to find out whether the shot works. For where Wan sits among other budget options, see Hailuo vs Wan for budget video and the best free AI video generators.
FAQ
Can I run Wan 2.5 without a GPU?
Yes, by using it online. Hosted on the 8frame canvas, Wan 2.5 runs server-side, so you generate 1080p clips in the browser with no local GPU. Running Wan 2.2 locally instead requires a capable card: at least 24 GB of VRAM for the documented 5B setup. Wan 2.5 itself has no public weights, so it cannot run locally at all.
How much VRAM does Wan need locally?
Per the Wan 2.2 model card, the TI2V-5B model's documented command needs a GPU with at least 24 GB of VRAM (an RTX 4090, for example) and takes under 9 minutes for a 5-second 720p clip; the 14B models run with offloading unless you have at least 80 GB. Community quantized builds go lower, but we have not tested them.
Is Wan 2.5 really free?
Locally, only for Wan 2.2: its weights are published, so there's no per-clip charge, but you pay in hardware and setup. Wan 2.5 has no public weights and no free route. On 8frame, which is subscription-based with no free tier, Wan 2.5 runs 102 credits a clip and Wan 2.2 runs 50 (from a start image or video only), so the $19/month Starter covers about 9 or 20 clips, watermark-free.
Is Wan good enough for client work?
We have not run a controlled quality comparison, so judge it on your own shots. Wan is priced for prototyping, mood tests and b-roll, and a common workflow is to draft on Wan, then rerun the winning shot on a model you trust for the final delivery.
Sources
- Wan 2.2 models, licence, VRAM and generation time: huggingface.co/Wan-AI/Wan2.2-TI2V-5B, checked 2026-10-01
- Published Wan weights (newest is the Wan 2.2 line): Wan-AI on Hugging Face, checked 2026-08-28
- 8frame prices, defaults and top-up packs: 8frame tool registry and pricing, read 2026-10-01
- Wan 2.2 input requirement on 8frame (start image or input video): 8frame video generation code, read 2026-10-01
No GPU, no problem. Generate with Wan online on the 8frame canvas, then finish the winners on another model in the same canvas. See the cheapest AI video generator math to plan the rest of your budget.
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.