Models

The model catalog is live and changes as providers add, retire, or update models. GET /v1/models lists the catalog; GET /v1/models/{model} returns a single model by ID or alias. Claude-compatible clients can use GET /v1/models_anthropic for an Anthropic-shaped catalog.

The catalog is the best way to decide which endpoint and request options to use. Inspect a model's input/output modalities, provider mappings, pricing, context window, and capability flags before sending traffic.

Examples

curl "https://api.llmapi.ai/v1/models"

curl "https://api.llmapi.ai/v1/models/claude-sonnet-4-6"

The list endpoint accepts optional lifecycle filters:

curl "https://api.llmapi.ai/v1/models?exclude_deprecated=true" \
  -H "Authorization: Bearer $LLM_API_KEY"

include_deactivated=true includes deactivated models for migration or diagnostics. exclude_deprecated=true removes models marked deprecated. The normal catalog omits deactivated entries and is suitable for model pickers.

Choosing a model

Catalog field	How to use it
`id` / aliases	Send the ID or alias in the request's `model` field.
`providers`	Shows available upstream providers and provider-specific model names.
`input_modalities`, `output_modalities`	Confirm whether the model accepts text, image, audio, video, or produces images.
Capability flags	Check streaming, tools, reasoning, JSON/schema output, embeddings, or batch support.
Pricing	Estimate cost before exposing a model to users. Pricing and availability can change.
Context and output limits	Keep prompts and requested output within the model's limits.

To force one provider, use the provider/model form supported by the gateway, for example google-vertex/gemini-2.5-pro. Without a provider prefix, LLM API routes according to the model's configured providers and fallback policy.

Anthropic-compatible catalog

GET /v1/models_anthropic returns a catalog shaped for clients that speak the Anthropic API, including Claude Code. It is a compatibility endpoint; use GET /v1/models for the complete LLM API catalog and capability metadata.

Use the catalog’s capability and modality fields to select models for vision, audio, image, video, embeddings, or batch workloads. The Models page also provides a browsable live catalog.