Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Zhipu AI / Z.ai / GLM-4

GLM-4-Plus

High-capability GLM-4 language model for text understanding, instructions and long-context work.

Language models (LLMs)
availableClosed weightsText generation
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
9 of 16 specification fields collected
Context window128KSee specification notes below
Maximum output4KSee specification notes below
Input billingSee ratesBigModel China
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
128KSource for Context / input limit (opens in a new tab)

Provider uses K notation; no exact integer is inferred.

Maximum output
4KSource for Maximum output (opens in a new tab)
Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
SupportedSource for Tool calling (opens in a new tab)
Structured output
SupportedSource for Structured output (opens in a new tab)

JSON-format output is documented; strict JSON Schema adherence is not assumed.

Streaming
SupportedSource for Streaming (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)

The guide includes an explicit GLM-4-Plus job example.

Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Not collected
Languages
Not collected
Context caching
SupportedSource for Context caching (opens in a new tab)

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

BigModel China

Provider source (opens in a new tab)
API model ID
glm-4-plus
Access
Provider account and authorized service access required. Public documentation reviewed; no paid inference performed.
Regions
Native open.bigmodel.cn service. Account eligibility and deployment region require provider confirmation; international Z.ai terms are not assumed.
API endpoint URL
POST https://open.bigmodel.cn/api/paas/v4/chat/completionsSource for API endpoint URL (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Text token inferenceInput1CN¥5.001M tokens
Text token inferenceOutput1CN¥5.001M tokens
Text token inferenceCached input1CN¥2.501M tokens
Batch inferenceBatch Input2CN¥2.501M tokens
Batch inferenceBatch Output2CN¥2.501M tokens

Rate conditions

  1. Standard GLM-4-Plus native API rate. Cache-hit input replaces the standard input charge for hit tokens. Published Chinese yuan rates, without currency conversion. Account quotas, contracts and separate service charges may apply.
  2. Supported asynchronous Batch API jobs receive 50% of standard pricing; this is not a cache discount. Published Chinese yuan rates, without currency conversion. Account quotas, contracts and separate service charges may apply.
Pricing source (opens in a new tab)Pricing source (opens in a new tab)

Data & enterprise terms

BigModel service terms sections 4.4 and 4.6 address customer data ownership and minimum necessary storage with anonymization and encryption. This is not a zero-retention or no-training guarantee.

Read the service policy (opens in a new tab)

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

BigModel China

glm-4-plus
Documented
Support evidence (opens in a new tab)
How it works
Implicit matching of repeated input prefixesSource for How it works (opens in a new tab)
Minimum reusable input
A repeated prefix above 500 tokens is recommended; a hard minimum is not specifiedSource for Minimum reusable input (opens in a new tab)
Lifetime & refresh
Not collected

No cache lifetime verified.

How to enable it
Automatic; no manual cache configuration requiredSource for How to enable it (opens in a new tab)
Usage fields
usage.prompt_tokens_details.cached_tokensSource for Usage fields (opens in a new tab)
Billing rules
Cache hits have the model-specific input rate. Cache storage is temporarily free; no permanent free-storage promise or end date is established.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Reuse identical prefix content; short prompts may not hit. Rates apply to the native BigModel endpoint, not every host.Source for Limits & data boundaries (opens in a new tab)
Cache reads, writes & storage
ChargeApplies toRate / unit
Cached input1Text token inferenceCN¥2.50 / 1M tokensPricing evidence (opens in a new tab)

Rate conditions

  1. Standard GLM-4-Plus native API rate. Cache-hit input replaces the standard input charge for hit tokens. Published Chinese yuan rates, without currency conversion. Account quotas, contracts and separate service charges may apply.

Observed 13 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 13 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Inferred from the exact named native model service in the official BigModel catalog and API or managed-service documentation, cross-checked against the developer’s public-weight inventory. No weights for this exact release were found in that inventory. A paid endpoint alone is not the basis: downloadable GLM releases and uncertain aliases are excluded pending exact mapping.

Classification evidence (opens in a new tab)

Sources reviewed