Replicate Model Deployments

Replicate documents running and deploying machine learning models via versioned containers with an HTTP API, integrating third-party apps to model versions hosted on Replicate.

• Select a model version on Replicate
• Create predictions via API or clients
• Optionally use deployments for dedicated capacity as documented

• Cold boots and hardware tiers affect latency
• Model licenses are defined by each model’s publisher

Replicate Model Deployments is tracked so teams can evaluate how AI surfaces connect to apps, data, and workflows using official documentation—not marketing claims.

• Replicate Docs
• Replicate HTTP API