Model IDs

The model field in every request is a model ID such as glm-5.3. IDs are lowercase vendor names with their dots kept.

These are examples. The current list, with prices, is in the model catalog. Copy the ID from the model page instead of typing it.

#IDs, not page URLs

A model page URL is readable but is not the ID. glm-5.3 lives at /models/glm-5-3: the dot became a dash. Always send the ID shown on the page, with its dots and in lowercase (deepseek-v4-pro, not DeepSeek-V4-Pro).

#Your key's group decides what you can call

Every API key belongs to a group, and the group decides which models the key can use. A model in the catalog may not be available to every group. List the models your key can call:

curl https://bigmodel.io/v1/models \
  -H "Authorization: Bearer $BIGMODEL_API_KEY"

import os
from openai import OpenAI

client = OpenAI(base_url="https://bigmodel.io/v1", api_key=os.environ["BIGMODEL_API_KEY"])
for model in client.models.list():
    print(model.id)

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://bigmodel.io/v1", apiKey: process.env.BIGMODEL_API_KEY });
for await (const model of client.models.list()) console.log(model.id);

The response is {"object": "list", "data": [...]} and each entry has an id. Add ?configured_only=true to list only models configured for the group. Some Chat Completions only groups return an empty list even though they serve models; in that case use the IDs from the catalog and test with one request.

GET /v1/models needs a positive balance. With a zero balance it returns 403 INSUFFICIENT_BALANCE.

#OpenRouter-style names

You can also send OpenRouter IDs such as z-ai/glm-5.3 or deepseek/deepseek-v4-pro:nitro. If the exact name is not served, the gateway tries it without the routing suffix and without the provider/ prefix. :free is never removed. The full rules are in OpenRouter compatibility.

Plain IDs without / or : are never rewritten.

#Model names in Usage

Always send the catalog ID. The Usage page may also show the provider's own name for the model that answered.

#What an unknown model returns

There is no "model not found" 404 for text models. A misspelled ID, or one your group does not serve, returns:

{"error":{"type":"api_error","message":"No available accounts: no available accounts supporting model: glm-5.3x"}}

with HTTP 503. Check the ID first: retrying a wrong ID will not help. If the ID is correct and listed for your key, check Status and retry later.

#Verify the setup

Run the cURL list command, pick an ID from the output, and send one request with it. In Usage, the request appears with that model.

#Troubleshooting

503 api_error "No available accounts: no available accounts supporting model: X". The ID is misspelled (check dots, case and dashes) or not served to your key's group. Copy it from the model page, or create a key in a group that serves the model.

403 INSUFFICIENT_BALANCE on /v1/models. The model list needs a positive balance. Top up in Billing.

/v1/models returns an empty data array. Some Chat Completions only groups do not publish a list. Use the catalog IDs and test with a request.

A tool says the model is not recognized. Tools such as Claude Code keep their own model lists. The request can still work; see the tool's guide, for example Claude Code.

A :free model fails. :free is part of the ID and is never removed. Use the exact ID shown in the catalog.

#Related

• OpenRouter compatibility for provider/model names and fallback lists.
• Models API reference.
• Model catalog and Pricing.