Models and Providers

Genfeed uses a provider registry. Model names, versions, capabilities, pricing, and availability can change independently of these docs, so the deployed model selector and API responses are the source of truth.

Provider adapters

The repository contains adapters for provider families including:

• OpenAI, Anthropic, OpenRouter, and xAI for language-model paths;
• Replicate and fal.ai for hosted media/model execution;
• ElevenLabs for voice/audio paths;
• ComfyUI for configured self-hosted workflows; and
• additional media providers such as Kling, Leonardo, HeyGen, Hedra, and Higgsfield.

An adapter’s presence does not guarantee that every model is enabled. Provider accounts, API keys, regional access, upstream approval, deployment mode, and the active registry determine what appears.

Content families

Registered models can support text, image, video, voice/audio, or transformation workflows. Check the selected model’s current input/output contract before building a workflow. Do not assume a fixed resolution, duration, format, or feature across models in the same family.

Billing boundary

• Local/BYOK execution is not metered by Genfeed; the configured provider can still charge or impose quotas.
• Managed Cloud execution uses the current credit and entitlement rules returned by the hosted product.
• This page intentionally does not publish fixed per-model credit costs or plan availability.

Programmatic use

Models are available through REST, workflow, CLI, or managed interfaces only when the corresponding route and registry entry exist in the deployed release. Asynchronous status, webhooks, or batch behavior is path-specific.

Verify a model path

1. Confirm the model appears in the deployed selector or registry response.
2. Confirm the required provider credential and account access.
3. Review input/output constraints and provider data terms.
4. Run a small, non-sensitive request.
5. Record the provider/model identifier used by the successful run.

For Community, see Provider Registry. For execution and billing boundaries, see Execution Boundaries.