GLM-5.3

GLM-5.3 for complex software engineering and long-horizon agent tasks

Published
scheduleReleasedAugust 14, 2026

GLM-5.3 supports a 1M-token context window and 128K maximum output with always-on reasoning at low, high or max effort. It is available in GLM Coding Plan, while the model API is not yet generally available.

starsCapabilities

codeFunction callingdata_objectStructured output

paymentsContext and pricing

Context limit1,000,000
Max output131,072
Input price$1.4/ 1M tokens
Output price$4.4/ 1M tokens
Cached input price$0.26/ 1M tokens

descriptionOverview

Overview

GLM-5.3 is Zhipu AI's latest flagship text model, using model ID glm-5.3. It shares the GLM-5.2 base model and improves complex software engineering, terminal work, cybersecurity and long-horizon agents through post-training. Official documentation lists a 1,048,576-token context window and 131,072-token maximum output.

Reasoning

Reasoning is always enabled. thinking.type: disabled is unsupported, while reasoning_effort accepts low, high or max and defaults to max.

Availability

GLM-5.3 is available through GLM Coding Plan. Zhipu states that the model API will launch later, so this wiki records official capabilities without publishing an unannounced API token price.

lightbulbUse cases

  • Complex software engineering
  • Long-horizon agents and terminal tasks
  • Cybersecurity analysis and vulnerability discovery
  • Large-context reasoning

thumb_upStrengths

  • 1M context and 128K maximum output
  • Stronger coding than GLM-5.2
  • Low, high and max reasoning effort
  • Optimized for long-running agents

infoLimitations

  • Model API is not yet generally available
  • Reasoning cannot be disabled
  • Text input only
  • Official API pricing is not yet published

linkReferences

This content is compiled from official documentation and public sources. Always refer to official documentation for final details