What is Graydient?
Graydient is GPU cloud infrastructure for open AI models. You get flat-rate, unlimited GPU rendering for open image, video and LLM models, hosting for the models you bring, and an API. There are no credits, tokens or boosts.
Graydient is not an app. It is the electricity and pipes underneath: data-center GPUs with open models preloaded, reachable from the API or from compatible apps made by independent developers.
Who it is for
- Creators who want to render as much as they like without watching a balance.
- App builders who need a render provider behind their product.
- Developers who want open models, workflows and LLMs over an API.
- Studios that want their own model library online without buying or maintaining GPUs.
What runs on it
- Images: text-to-image, upscaling and face swap (subject to the acceptable use policy).
- Video: text-to-video, photo-to-video, extending and upscaling videos.
- Music and dubbing: unlimited on Video Plus and Video Max.
- LLMs: unlimited open LLMs, plus custom bots.
Preloaded at the time of writing: KREA 2 (default image model), MiniMax H3 (default video model), Grok (default LLM), plus Wan 2.2, Hunyuan, Flux 2, Anima, Ernie, SDXL and others.
GPUs and speed
Graydient owns or leases A100, A6000 and RTX 5090 GPUs, upgrades them regularly and does not partition them. Servers run current AMD and Intel CPUs with 250GB or more of RAM and NVMe drives.
Typical times: images, upscales and face swaps in about 30 seconds, videos in about 60 seconds and photo-to-video in about 150 seconds. Each job gets a full GPU for up to 3 minutes (Images, Video Plus) or up to 6 minutes (Video Max).
Bring your own models
Every plan lets you install your own LoRAs and models: one-click installs from the ModelTrends catalog, open weights from Hugging Face, community .safetensors files or your own uploads. Most .safetensors files under 25GB work, apart from a few edge cases. Read more about how to host your own AI models.
Ways to use it
- The API (Video Max only): a personal API for open models, LLMs and ComfyAPI API workflows. See the AI rendering API.
- Compatible apps from independent developers: some third-party apps let you pick Graydient as their render provider, for example BitVector's easy chat creator (Prism), pro studio (SpyGlass) and Discord bot (Vector), and PirateDiffusion. These are independent products, not Graydient apps; their own sites explain how to use them.
Pricing
One flat fee in US dollars, billed yearly or monthly. You can upgrade, downgrade or cancel anytime.
| Plan | Yearly (per month) | Monthly | Includes |
|---|---|---|---|
| Images | $25$299.99 billed yearly | $29.99 | Unlimited images, your own LoRAs, custom bots, unlimited LLMs, 500GB storage. No video, no API. |
| Video Plus | $41.67$499.99 billed yearly | $49.99 | Everything in Images, plus short videos (best at 480p), unlimited music and dubbing, video extend and upscale, 1TB storage. No API. |
| Video Max | $83.25$999 billed yearly | $99.99 | Unlimited videos, longer videos up to 1440p, bigger image batches, fastest render times, 5TB storage, personal API. |
Trust and billing
- Clear flat pricing. No credits, tokens, boosts or hidden fees. Storage is the only cap: when your drive is full, delete files or change plans. You can download your work at any time.
- Cancel anytime in the billing portal.
- Upgrade or downgrade anytime in the same portal. Charges are prorated.
- Forgot your password? Use "forgot password" on the portal's login page.
- 7-day money-back guarantee, no questions asked. There is no free trial: if Graydient is not a good fit, ask for a refund.
- Quick refunds. Refunds are processed promptly and usually arrive in 3 to 4 business days in the US, longer elsewhere, depending on your bank.
- Privacy. Logins are protected with 2FA. Graydient does not train models, and the photos you upload are not sold or used for training. See the privacy policy.
Questions
Is it really unlimited?
Yes, within your plan's features. There are no credits or boosts. Each job gets a set amount of full-GPU time, and you can send the next one when it finishes.
Is Graydient an app?
No. It is infrastructure. Use the API, or a compatible app from an independent developer.
Which plan includes the API?
Only Video Max, and only for one human user. Automated traffic is not allowed on any personal plan: no bots, scripts, batch jobs or resold access. You can use the personal API to build and test your own app; going live with automation or a multi-user product needs the Business plan at business.graydient.ai. See the AI rendering API.
How do I cancel or change my plan?
Log in to the billing portal. There you can cancel anytime, or upgrade or downgrade with prorated charges. If you have forgotten your password, use "forgot password" on that page.
Can I try it for free?
Not right now, but every plan comes with the 7-day money-back guarantee.
What is the difference between Video Plus and Video Max?
GPU time per job: up to 3 minutes on Video Plus and up to 6 on Video Max. Max supports longer videos up to 1440p, bigger batches and includes the API. For the full cost picture, read flat-rate vs. credits.