Models
LLM API's model catalog changes as providers ship new models and retire old ones, so this page describes how the catalog is structured rather than freezing a specific list or count. For the exact models available to your key right now, call the live endpoint:
curl https://api.llmapi.ai/v1/models \
-H "Authorization: Bearer $LLM_API_KEY"
INFO
Full request/response schema for GET /v1/models is in API Reference → Models.
Model catalog
INFO
This table is loaded live from GET /v1/models each time the page opens, so it always reflects
the current catalog. You can also browse it in the
dashboard.
The live model catalog is available from GET https://api.llmapi.ai/v1/models.
Supported providers
LLM API routes to models from the following provider families:
| Brand |
Provider family |
| Claude |
Anthropic |
| GPT |
OpenAI, Azure OpenAI |
| Gemini |
Google, Google AI Studio |
| Grok |
xAI |
| Kimi |
Moonshot |
| DeepSeek |
DeepSeek |
| Llama |
Meta |
| Qwen |
Alibaba (Qwen) |
| GLM |
Zhipu AI (GLM/Z.ai) |
| Mistral |
Mistral |
| ElevenLabs |
ElevenLabs (audio) |
| MiniMax |
MiniMax |
| Cohere |
Cohere |
| Cartesia |
Cartesia (audio) |
| Deepgram |
Deepgram (audio) |
| AssemblyAI |
AssemblyAI (audio) |
| Amazon |
Amazon (Bedrock-hosted models) |
Some brands are served by more than one upstream provider (for example GPT models via both OpenAI
directly and Azure OpenAI) — see [Provider routing](/features/routing) for how LLM API picks between
them, and [Custom Providers](/features/custom-providers) for bringing your own provider credentials.
What's on a model
Each entry in the catalog carries:
Identity — id, name, family (the provider grouping above), aliases, kind, released_at, and deprecated_at / deactivated_at where they apply.
Capabilities — per-provider flags such as streaming, vision, tools (function calling), reasoning, json_output, structured_outputs, audio_input, web_search and supportsBatch.
Modalities — architecture.input_modalities and architecture.output_modalities list what the model accepts and produces.
Pricing & limits — model-level pricing.prompt / pricing.completion quote USD per token; multiply by 1,000,000 for the per-1M figures shown above. Because one model can be served by several upstream providers at different rates, each entry in providers[] carries its own pricing and contextSize.
Filtering the catalog
GET /v1/models takes no query parameters — it returns the whole catalog in a single response, and you filter client-side on the fields above. The table on this page does exactly that.
To narrow by capability, read the flags on each entry's providers[]; to narrow by modality, read architecture.output_modalities. The dashboard's model browser offers the same filters as a UI.