Skip to content

Instantly share code, notes, and snippets.

@apnea
Created May 10, 2026 10:58
Show Gist options
  • Select an option

  • Save apnea/51bf344b7b047459e151b1ba97d23eb9 to your computer and use it in GitHub Desktop.

Select an option

Save apnea/51bf344b7b047459e151b1ba97d23eb9 to your computer and use it in GitHub Desktop.
Z.AI Coding Plan — OpenCode agent-to-model mapping

Z.AI Coding Plan — OpenCode Agent Mapping

Quota Cost per Model

Peak hours: 14:00–18:00 UTC+8. Off-peak: all other times. Monthly quota is equivalent to ~15–30× the subscription fee, converted at API pricing rates.

Model Quota (Peak: 14:00-18:00 UTC+8) Quota (Off-Peak) Temporary (thru June)
GLM-5.1 1× off-peak
GLM-5-Turbo 1× off-peak
GLM-4.7
GLM-4.5-Air

Agent → Model Mapping

Principle: use the flagship model (GLM-5.1) where reasoning quality directly impacts output, and the 1× models (GLM-4.7, GLM-4.5-Air) for routine or lightweight work to conserve quota.

Agent Model Rationale Quota
build GLM-5.1 Primary agent — flagship for complex coding tasks 2-3×
plan GLM-5.1 Deepest reasoning needed for planning complex changes 2-3×
general GLM-4.7 Subagent multi-step tasks — strong coding/reasoning at 1×
explore GLM-4.7 Code exploration — good understanding, 1× quota
compaction GLM-5.1 Critical for context quality — needs best reasoning 2-3×
title GLM-4.5-Air Simple task, lightweight sufficient
summary GLM-4.5-Air Simple task, lightweight sufficient

Thinking Mode

  • Enabled (default) on build, plan, general, explore, compaction — these chain multiple tool calls and benefit from interleaved reasoning between steps.
  • Disabled on title and summary — trivial tasks that don't need reasoning, disabling saves quota and latency.

Key Sources

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment