Replicate in AGNT
Run community models — vision, audio, image — by API. Every AGNT workflow is provider-portable, so the same agents run here or anywhere else you point them.
Why teams pick it
Models
- thousands of hosted models
Strengths
- long-tail model access
- image and audio generation
- pay-per-run
In AGNT
Agents, workflows and tools use Replicate like any other provider — same nodes, same receipts, swap models per step if you want.
The long tail of models
Replicate’s catalogue reaches far beyond text: image generation, audio, video, vision, and thousands of community checkpoints. For agent workflows that need a specialised capability once — transcribe this, segment that, generate an illustration — pay-per-run access removes the need to host anything yourself.
What to watch
Cold starts are real, so a rarely-used model may take noticeably longer on first call. Community models vary in quality and maintenance, which makes pinning a specific version important if a workflow depends on consistent behaviour.
Using Replicate in a workflow
The practical question is not whether Replicate is good but which steps deserve it. Point thousands of hosted models at the judgment calls — the places where long-tail model access changes the answer — and route the high-volume routine work elsewhere in the same workflow. Swapping either side later does not touch the workflow itself.
Connect Replicate in two minutes
- Download AGNT Community Core — free, local-first, no account needed to run.
- Open Settings → Providers, choose Replicate, and paste your API key. Connect by pasting an API key into AGNT’s vault — stored encrypted on your machine, never uploaded.
- Give an agent a job — or install one from the marketplace — and watch the first receipt come back.
Replicate in AGNT — common questions
What is Replicate best for in a workflow?
Specialised, occasional capabilities — image, audio and vision steps that do not justify hosting a model.
Why is the first call slow?
A cold model has to load. Frequently used models stay warm; rarely used ones will not.
Should I pin model versions?
Yes, if consistent output matters. Community models can change underneath you otherwise.