ReplicateServerless cloud GPU inference platform for AI models; unified API to run hundreds of open-source models (text-to-image, image-to-video, LLMs, speech, video, restoration, etc.); supports one-click custom model deployment and auto-scaling
Free Tier
- Quota
- Curated models in the try-for-free collection come with a fixed number of free runs, not free forever (No. The models in this collection are free for a limited number of runs); after the free use limit you must add billing and purchase credit to continue.
- Cycle
- One-time
Replicate's curated free collection: popular models for image/video generation and photo restoration are free to try within limits; other models are pay-per-use
Similar free resources
- NVIDIA NIMFree: up to 40 requests/min for most models, no per-token billing; personal limit shown in dashboard
- iFlytek Xingchen MaaSSome models are time-limited free: Spark-X2.5-4B (native 1M token context), DeepSeek-OCR.
- OpenRouter:free models: 20 req/min; 50/day before top-up, 1,000/day after $10 purchased
- PerchanceFree access to numerous AI generators; many AI image, chat, and story generators require no sign-up and have unlimited use.
About this data：Last checked · Added · Source replicate.com