GLM-4.6 Air: Is It a Real Model? Specs, API Access, and Pricing

Short answer: The Z.ai documentation reviewed for this guide documents GLM-4.6 and GLM-4.5-Air, but does not confirm a model named GLM-4.6-Air. That does not prove GLM-4.6-Air does not exist; its identity, specifications, pricing, and API availability remain unconfirmed by the sources reviewed. Don’t assume details for either documented model apply to GLM-4.6-Air.

GLM-4.6 Air vs. GLM-4.6 vs. GLM-4.5-Air

Name	What Z.ai’s documentation establishes	What that means for selection
GLM-4.6-Air	The sources reviewed do not confirm this exact model name.	Treat availability and capabilities as unverified. Don’t use a guessed model ID in production.
GLM-4.6	Z.ai documents a 200K-token context window, up to 128K output tokens, text input and output, and the API model ID glm-4.6.	Consider the documented model if its limits and supported features fit your workload. These are GLM-4.6 specifications—not Air specifications.
GLM-4.5-Air	Z.ai documents a 128K-token context window, up to 96K output tokens, and a more streamlined 106B-parameter design.	It is a separately documented Air model. Its specifications and prices do not establish GLM-4.6-Air specifications or prices.

Z.ai describes GLM-4.6 as supporting advanced reasoning and tool use during inference, with stronger performance in coding and tool-using agents. Its published specifications identify text input and text output. These statements concern GLM-4.6, not GLM-4.6-Air. Don’t infer image input from the similar name GLM-4.6V, which is a separately documented vision model. (GLM-4.6 documentation)

Z.ai describes GLM-4.5-Air as a more streamlined model, with 106B total parameters and 12B active parameters; its documentation gives 128K context and 96K maximum output. Those details apply to GLM-4.5-Air, not GLM-4.6-Air.

Pricing reference: check current rates before budgeting

Z.ai’s pricing page labels its model rates in USD per 1 million tokens, with separate columns for Input, Cached Input, Cached Input Storage, and Output. In the listed rows, GLM-4.6 is shown at $0.60 input, $0.11 cached input, limited-time free cached-input storage, and $2.20 output. GLM-4.5-Air is shown at $0.20 input, $0.03 cached input, limited-time free cached-input storage, and $1.10 output. These figures describe the named models in Z.ai’s pricing table, not GLM-4.6-Air or TokenForge. Rates and promotional terms can change, so confirm the applicable current prices and conditions with the provider before budgeting. (Z.ai pricing documentation)

Illustrative estimate using the listed standard input and output rates: For 1 million uncached input tokens and 250,000 output tokens, the model-token charges would be:

• GLM-4.6: (1 × $0.60) + (0.25 × $2.20) = $1.15
• GLM-4.5-Air: (1 × $0.20) + (0.25 × $1.10) = $0.475

This is arithmetic using the rates listed above, not a quote or measured invoice. It excludes cached-input treatment and any other applicable charges. A lower listed token price alone does not establish comparable task quality; compare output quality, context needs, latency, and tool behavior on representative tasks.

Can a production API serve GLM-4.6 Air?

The documentation reviewed confirms that Z.ai documents the API model ID glm-4.6; that supports access to GLM-4.6 through Z.ai’s documented API, not access to GLM-4.6-Air. The evidence reviewed for this guide does not establish TokenForge support for GLM-4.6-Air. Confirm the exact model ID in the current catalog of whichever provider you intend to use. (Z.ai migration guide)

For Z.ai’s documented chat-completions API, the reference shows the route https://api.z.ai/api/paas/v4/chat/completions, Bearer-token authentication, and a request with model and messages fields. The example below uses the documented GLM-4.6 model ID; it is not a GLM-4.6-Air request and has not been live-tested here. (Z.ai chat-completions reference)

curl --request POST \
  --url https://api.z.ai/api/paas/v4/chat/completions \
  --header "Authorization: Bearer $ZAI_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "model": "glm-4.6",
    "messages": [
      {"role": "user", "content": "Summarize this test request."}
    ]
  }'

Before building around a model, verify its exact model ID in your provider’s current catalog, along with the endpoint, authentication, features your application requires, applicable usage limits, and price unit. A model name in an article or coding-tool menu does not confirm that it is enabled on your production API account.

Which model should you choose?

• If you need the documented GLM-4.6 model: Evaluate glm-4.6 where your provider explicitly lists it. Z.ai documents 200K context and up to 128K output; test your own tasks rather than assuming those limits guarantee better results for your workload. (GLM-4.6 documentation)
• If you want a documented Air model: Compare GLM-4.5-Air using its own specifications and the intended provider’s current pricing. Do not treat its documentation as evidence about GLM-4.6-Air.
• If your requirement specifically says “GLM-4.6-Air”: Ask the intended provider to confirm the canonical model name and ID, and request model-specific capability and pricing documentation before using it. Don’t silently substitute GLM-4.6 or GLM-4.5-Air.

Bottom line: GLM-4.6 and GLM-4.5-Air are distinct documented options. The sources reviewed do not confirm GLM-4.6-Air. Choose based on the exact model your API provider supports, its documented limits and current rates, and its results on a representative evaluation set—not on a similar model name.

Sources

GLM-4.6 — Z.AI Developer Documentation

GLM-4.5 — Z.AI Developer Documentation

Pricing — Z.AI Developer Documentation

Migrate to GLM-4.6 — Z.AI Developer Documentation

Chat Completion — Z.AI Developer Documentation