AIsa Model Catalog: Model IDs, Routes, and Capabilities

Find current AIsa model IDs, supported endpoints, and capabilities via GET /v1/models or the live catalog at aisa.one/models.

AIsa routes LLM, image, video, and embedding models through a single API key. Because models are frequently added and retired, this page explains how to query the catalog dynamically rather than maintaining a static list.

GET /v1/models

Machine-readable list of model IDs, routes, and capability tags for applications and agents.

aisa.one/models

Browse, filter, and compare models, with current prices.

List models with the API

Send your API key as a Bearer token. Listing models is free; requests without an API key return 402.

curl https://api.aisa.one/v1/models \
  -H "Authorization: Bearer $AISA_API_KEY"

The response follows the standard OpenAI list format, extended with an aisa metadata object for each model:

{
  "object": "list",
  "data": [
    {
      "id": "claude-fable-5-1",
      "object": "model",
      "created": 1788315984,
      "owned_by": "Anthropic",
      "aisa": {
        "display_name": "claude-fable-5-1",
        "model_type": "text",
        "capabilities": ["chat", "long-context", "reasoning", "tools"],
        "preferred_route": "anthropic",
        "routes": [
          { "key": "openai", "path": "/v1/chat/completions", "method": "POST" },
          { "key": "anthropic", "path": "/v1/messages", "method": "POST", "preferred": true }
        ]
      }
    }
  ]
}

Field	Meaning
id	The exact string to pass as model.
owned_by	Upstream provider, such as OpenAI, Anthropic, or Alibaba.
aisa.model_type	text, image, video, or embedding.
aisa.capabilities	Capability tags, such as chat, tools, reasoning, vision, image, videos, or embeddings.
aisa.routes	Every AIsa endpoint serving this model, specified as method + path.
aisa.preferred_route	Present on some models. The key of the recommended route (which also has "preferred": true).

Filter the list with jq:

# Models served by the Anthropic-compatible /v1/messages endpoint
curl -s https://api.aisa.one/v1/models -H "Authorization: Bearer $AISA_API_KEY" \
  | jq -r '.data[] | select(any(.aisa.routes[]; .path == "/v1/messages")) | .id'

# Image models
curl -s https://api.aisa.one/v1/models -H "Authorization: Bearer $AISA_API_KEY" \
  | jq -r '.data[] | select(.aisa.model_type == "image") | .id'

Endpoints by model type

Model type	Endpoints	Notes
Text	POST /v1/chat/completions, POST /v1/messages, POST /v1/responses	Most text models serve both OpenAI- and Anthropic-compatible routes. Check routes before switching protocols.
Image	POST /v1/images/generations, POST /v1/images/edits	Not every image model supports edits.
Video	POST /v1/video/generations, then GET /v1/video/generations/{task_id}	Asynchronous: submit a job, then poll for the result.
Embedding	POST /v1/embeddings

Calling a model on an unsupported route returns 400 with error.code set to model_route_not_supported; error.supported_routes lists the routes that work. Request and response schemas for each endpoint are documented in the API Reference.

Call a model

Point any OpenAI-compatible SDK at https://api.aisa.one/v1 and pass a model id from the catalog:

from openai import OpenAI

client = OpenAI(api_key="YOUR_AISA_API_KEY", base_url="https://api.aisa.one/v1")

response = client.chat.completions.create(
    model="qwen3.7-max",
    messages=[{"role": "user", "content": "Compare these model options for a coding agent."}],
)
print(response.choices[0].message.content)

For the Anthropic SDK and other clients, see OpenAI / Anthropic SDK.

Browse and compare on aisa.one

For interactive browsing, visit aisa.one/models. You can search by model, provider, or endpoint, filter by provider and capability, and compare models side by side. It also displays each model’s current price, which these docs do not list.

How models are billed

Models draw on the same prepaid balance as data APIs. The billing unit depends on the model:

• Text and embedding models: Billed per token, with separate rates for input and output, and for cache reads or writes where caching is supported.
• Image models: Billed per token, per request, or per image (sometimes varying by image size), depending on the model.
• Video models: Billed per output second or per token, by resolution.

The live catalog shows each model’s billing type and current price. The final charge for each call is recorded in your usage logs. See Billing & Credits for balances, spend limits, and quotes.

Chinese models

Qwen, DeepSeek, Kimi, GLM, ByteDance Seed, MiniMax, and other Chinese models use the same endpoints. See Chinese AI Models for per-family guides.

Notes for agents

• Use only model IDs dynamically returned by GET /v1/models. Do not guess model IDs or reuse them from memory.
• Select the endpoint from the model’s routes list; prioritize the route marked "preferred": true when present.
• Do not assume a model supports a modality simply because its upstream family does. Always check model_type and capabilities.
• Listing models via GET /v1/models is free and allowed before the user approves any spend. Current prices are listed on aisa.one/models, not in these docs.