Back to all plans
天翼云
Dual-model access
China Telecom's coding Token Plan — token-quota billing, led by GLM-5 and DeepSeek-V3.2
Token PlanCoding Token Plan
SubscriptionToken API
Core models
GLM-5（正式版）DeepSeek-V3.2（旗舰版）
GLM-5（正式版）
Zhipu GLM-5 formal—all five coding Token Plan tiers; for complex reasoning and agent-style coding.
DeepSeek-V3.2（旗舰版）
DeepSeek V3.2 flagship on all five tiers—value mainline for codegen and daily engineering.
Plan details
Usage
25M tokens per month is better for first-time AI coding with light scripts, simple Q&A, and small edits rather than sustained heavy development.
Models
You can choose between GLM-5 and DeepSeek-V3.2 to experience stylistic differences at lower cost before deciding whether to upgrade allowance.
Highlights
It suits users trying AI coding tools for the first time and wanting a low-cost trial of the telecom official Token Plan.
If you already know you will frequently use the flagship model, this tier will feel tight quickly and higher tiers are more appropriate.
If you already know you will frequently use the flagship model, this tier will feel tight quickly and higher tiers are more appropriate.
Best for
Entry users, AI coding beginners, and individuals wanting a low-cost trial
Usage
80M tokens per month supports a steadier daily development rhythm, fitting regular workflow use rather than occasional trial.
Models
GLM-5 and DeepSeek-V3.2 remain selectable, letting you make more practical daily tradeoffs between cost and capability.
Highlights
It fits daily development, script maintenance, medium-complexity tasks, and people who will rely on AI coding lightly to moderately over time.
If project scale grows, contexts lengthen, and tasks become more complex, the 180M tier is more appropriate than forcing this one.
If project scale grows, contexts lengthen, and tasks become more complex, the 180M tier is more appropriate than forcing this one.
Best for
Daily developers, script maintainers, and developers with stable light-to-medium AI coding reliance
Usage
180M tokens per month works well as the main tier for most developers, covering more complex projects, sustained iteration, and moderately high engineering load.
Models
You can still switch between GLM-5 and DeepSeek-V3.2, but flagship usage should be controlled or headline allowance is consumed quickly.
Highlights
It suits advanced developers, long-running project maintainers, and mid-to-high-frequency users who want steadier token budgets.
This is the most balanced long-term primary choice in CtCloud Token Plan unless you are clearly in extreme usage territory.
This is the most balanced long-term primary choice in CtCloud Token Plan unless you are clearly in extreme usage territory.
Best for
Advanced developers, long-term maintainers, and mid-to-high-frequency primary AI coding users
Usage
380M tokens per month targets more complex projects and higher-frequency engineering work for teams or individuals deeply embedding AI coding into daily flow.
Models
GLM-5 and DeepSeek-V3.2 remain available, but this tier is better for using the flagship on key tasks rather than defaulting every request to it.
Highlights
It suits professional developers, complex project teams, and organizations needing a larger token pool to reduce switching anxiety.
If you are not yet at the extreme intensity of the 680M tier, this is usually enough for formal production pacing.
If you are not yet at the extreme intensity of the 680M tier, this is usually enough for formal production pacing.
Best for
Professional developers, complex project teams, and mid-to-high-frequency production users
Usage
680M tokens per month is clearly aimed at high-frequency AI developers, fitting sustained complex iteration, long-context work, and heavy daily generation.
Models
GLM-5 and DeepSeek-V3.2 remain available, but this tier works best as a unified quota pool for long-running high-frequency production environments.
Highlights
It suits high-frequency AI developers, multi-project teams, and organizations treating model assistance as core infrastructure.
It is obvious overkill for ordinary users; it only becomes worthwhile when you truly consume large token volumes every day.
It is obvious overkill for ordinary users; it only becomes worthwhile when you truly consume large token volumes every day.
Best for
High-frequency AI developers, multi-project teams, and organizations with extreme engineering workflows
Quota
Product type
Token Plan
Official product line
Coding Token Plan
5-hour limit
—
Weekly limit
—
Monthly limit
25M–680M tokens/mo (by tier)
Notes
- All five coding Token Plan tiers bill monthly (¥29/89/199/399/699)—no quarterly or annual discount tiers on the official site; detail pages show the main monthly rate only.
- Former GLM Coding Plan (Lite/Pro/Max ¥49/149/469 monthly) is sold out—choose the five coding Token Plan tiers.
- All five tiers support GLM-5 (formal) and DeepSeek-V3.2 (flagship)—one active plan per main account across plan types.
- Effective immediately on purchase; upgrade/renewal coming soon—repurchase after expiry; no refund except as required by law.
- Coding tools only—not a substitute for pay-as-you-go Token Service API; self-built backends use the API path.
Supported coding tools
OpenClawClaude CodeOpenCodeCursorClineChatboxCodeBuddyTRAEHermes AgentCodex CLI
Pricing and model data sourced from official vendor websites
FAQ
General9 条