AI Coding Plan Comparison - China and global coding subscriptions

Compare China and global AI coding plans in one place, including Zhipu, MiniMax, Kimi, Claude, Cursor, GitHub Copilot, ChatGPT, Gemini Code Assist and Windsurf.

2026 · China Coding Plan Watch

China AI Coding Plans

Zhipu GLM, MiniMax, Kimi, Volcengine Ark, Alibaba Bailian, Tencent Cloud, Baiyun Zhisuan and Xiaomi MiMo in one place. Compare price, models and usage limits before you subscribe.

8platforms tracked
¥29lowest monthly price
10+models covered
20+coding tools

Quick comparison

Start with price, models, quotas and standout features before opening the full detail pages.

ProviderEntry pricePromoCore modelsUsageTiersHighlight
Zhipu GLM¥49/mo—GLM-5, GLM-4.7, GLM-4.6~80 prompts / 5h3 tiersMCP included
MiniMax¥49/moNew Ultra ¥469MiniMax-M3 · M2.7Unified multimodal pool3 active tiersMultimodal Agent
Kimi¥49/mo—Kimi K2.5300-1200 API calls / 5h2 tiersMember perks
Volcengine Ark¥40/mo¥8.91Doubao, DeepSeek, Kimi, GLM and 6 models totalSeveral times Claude Pro2 tiersMost model choice
Alibaba Bailian¥40/mo¥7.9Qwen series + GLM + Kimi1200 req / 5h2 tiersLowest first month
Tencent Cloud¥40/mo¥7.9 first month · ¥20 second monthHY 2.0 · GLM-5 · Kimi-K2.5 · M2.51200 req / 5h2 tiersTencent in-house
Baiyun ZhisuanUsage-based¥450 creditsMiniMax-M2.5 · GLM-5 · DeepSeek · Kimi · QwenPer-token billingUsage-based¥450 bonus credits
Xiaomi MiMo¥39/mo12% first purchase offMiMo-V2.5-Pro · MiMo-V2.5 · TTS · V2 Pro/Omni60M-1600M Credits/mo4 tiersNo 5-hour cap

Plan breakdowns

Expand each homepage summary into a real plan card, then open the provider page for the complete view.

Alibaba Bailian Coding Plan

¥7.9 first month

Qwen plus third-party models, discounted first month and 50% off the second renewal month.

Models
  • qwen3.5-plus
  • qwen3-max
  • qwen3-coder-next
  • qwen3-coder-plus
  • glm-4.7
  • kimi-k2.5
Litepromo
1,200 requests / 5h9,000 / week18,000 / month
¥40/mo¥7.9 first month · ¥20 second month
Proadvanced
6,000 requests / 5h45,000 / week90,000 / month
¥200/mo¥39.9 first month · ¥100 second month

Zhipu GLM Coding Plan

MCP included

GLM-5 ranks in the top SWE-bench Verified tier, with quarterly and yearly discounts.

Models
  • GLM-5
  • GLM-4.7
  • GLM-4.6
Litelight
~80 prompts / 5h · ~400 / week100 MCP reads/searches monthlyGLM-4.7 / GLM-4.6
¥49/mo¥44/mo quarterly
Propopular
~400 prompts / 5h · ~2,000 / week1,000 MCP reads/searches monthlyGLM-5 / GLM-4.7 / GLM-4.6
¥149/mo¥134/mo quarterly
Maxhigh quota
~1,600 prompts / 5h · ~8,000 / week4,000 MCP reads/searches monthlyPriority access to new models
¥469/mo¥422/mo quarterly

MiniMax Token Plan

100+ TPS

Budget entry price with optional high-speed tiers and annual savings.

Models
  • M2.7
  • M2.7-highspeed
Starterentry
100 prompts / 5hentry-level coding use
¥29/mo
Plus
300 prompts / 5h3x Starter quota
¥49/mo
Max
1000 prompts / 5h10x Starter quota
¥119/mo
Plus High-speed
300 prompts / 5haround 100 TPS
¥98/mo
Max High-speedbest value
1000 prompts / 5haround 100 TPS
¥199/mo
Ultra High-speedheavy use
2000 prompts / 5h20x Starter quota
¥399/mo

Kimi Code Plan

Member perks

K2.5 across the board, 3x quota campaign messaging and extra member features.

Models
  • Kimi K2.5
Andantebasic
300-1,200 API calls / 5hrefreshes every 7 daysearly access to flagship model
¥49/mo¥39/mo yearly (¥468/year)
Moderatorecommended
larger quota every 7 daysmulti-device shared accessClaude Code / Roo Code support
¥99/mo¥79/mo yearly (¥948/year)

Tencent Cloud TokenHub Coding Plan

¥7.9 first month

Tencent Hunyuan plus top third-party models with staged promo pricing.

Models
  • Tencent HY 2.0 Instruct
  • GLM-5
  • Kimi-K2.5
  • MiniMax-M2.5
Litepromo
~1,200 requests / 5h~9,000 / week · ~18,000 / monthall supported models included
¥7.9 first month¥20 second month · ¥40 from month three
Propro choice
~6,000 requests / 5h~45,000 / week · ~90,000 / month5x Lite quota
¥39.9 first month¥100 second month · ¥200 from month three

Volcengine Ark Coding Plan

¥8.91 first month

The broadest model lineup with Auto mode and seasonal subsidy pricing.

Models
  • Doubao-Seed-2.0-Code
  • Doubao-Seed-Code
  • Kimi-K2.5
  • Kimi-K2
  • GLM-4.7
  • DeepSeek-V3.2
Lite75% off first month
several times Claude Pro usagemanual model switchingAuto routing mode
¥40/mo¥8.91 first month
Promost popular
5x Lite quotafor large-scale coding workseveral times Claude Max usage
¥200/mo¥44.91 first month

Baiyun Zhisuan Model API

¥450 credits

Register and verify to get ¥150, then earn another ¥300 after the first API call.

Models
  • MiniMax-M2.5
  • GLM-5
  • GLM-4.7
  • DeepSeek
  • Kimi
  • Qwen
Registration + verificationinstant
issued automatically after verificationcan offset API usage
¥150 credit
First API callbonus
triggered after first successful API callusually arrives in about 3 business days
¥300 credit

Xiaomi MiMo Token Plan

from ¥39/mo

Four-tier Credits subscription covering MiMo-V2.5-Pro, MiMo-V2.5, TTS and V2-series models.

Models
  • MiMo-V2.6-Pro
  • MiMo-V2.6-Flash
  • MiMo-V2.5-Pro
  • MiMo-TTS
Liteentry
60M Credits monthly~120 medium-complex tasks
¥39/mo$6/mo · ¥411.84/year
Standarddaily
200M Credits monthly~400 medium-complex tasks
¥99/mo$16/mo · ¥1045.44/year
Proprofessional
700M Credits monthly~1400 medium-complex tasks
¥329/mo$50/mo · ¥3474.24/year
Maxheavy use
1600M Credits monthly~3200 medium-complex tasks
¥659/mo$100/mo · ¥6959.04/year

Global AI coding plan comparison

Based on public pricing pages from official provider sites, with a focus on entry price, model access, team plans and quota framing.

ProviderEntry pricePromoCore modelsUsageTiersHighlight
Claude$20/moMax from $100Claude Sonnet / Opus / Claude CodeMax offers 5x or 20x5 tiersOfficial Claude Code
Cursor$20/moUltra $200Claude · GPT · GeminiPro+ 3x · Ultra 20x5 tiersIDE-native
GitHub Copilot$10/moNew Max $100 · Free availableGPT-5.4 · Claude · GeminiGitHub AI Credits unified model5 tiersGitHub ecosystem
ChatGPT$20/moPro $200GPT-5.4 · CodexBusiness can add Codex seats5 tiersChatGPT + Codex
Gemini Code AssistFreeFrom $19/seat/moGemini 2.5 · Code AssistIndividuals plan is free3 tiersGoogle stack
Windsurf$15/moTeams $30Cascade + major model mixPro 500 credits / month4 tiersAgent workflow

Global provider breakdowns

Expand Claude, Cursor, GitHub Copilot, ChatGPT, Gemini Code Assist and Windsurf into actionable plan summaries.

Claude Plans

Official Claude Code

Pro includes Claude Code, Max raises usage to 5x or 20x, and team plans scale up from there.

Models
  • Claude Sonnet
  • Claude Opus
  • Claude Code
Proindividual
Claude Code includedMore usageBest starting point
$20/mo$17/mo annual equivalent
Max 5x
5x Pro usagePriority accessHigher output limits
$100/mo
Team Standardteam
Central adminCollaboration featuresSeat-based billing
$25/seat/moannual billing

Cursor Pricing

IDE-native

The ladder from Hobby to Ultra and Teams is unusually easy to reason about.

Models
  • Claude
  • GPT
  • Gemini
Procore
Personal planUnlimited included modelsPaid usage bands above that
$20/mo
Pro+
3x OpenAI / Claude / Gemini usageMore headroomHeavy solo use
$60/mo
Teamsteam
Shared rulesOrg billingSSO and analytics
$40/seat/mo

GitHub Copilot Plans

GitHub ecosystem

Free, Pro, Pro+, Max and Business use unified GitHub AI Credits across Chat, Agent mode and CLI.

Models
  • GPT-5.4
  • GPT-5.4 mini
  • Claude
  • Gemini
Freeentry
2000 completions / month50 chat or agent requestsBest for validation
$0
Pro
300 premium requests / monthUnlimited included modelsPopular solo tier
$10/mo
Businessteam
Org policiesCentral management300 premium requests / user / month
$19/seat/mo

ChatGPT Plans

ChatGPT + Codex

Plus targets individuals, Pro targets very heavy use, and Business adds teamwork plus Codex seat assignment.

Models
  • GPT-5.4
  • GPT-5.4 mini
  • Codex
Plusindividual
More GPT-5.4 accessAppsCodex available
$20/mo
Pro
Expanded CodexHigher model limitsAdvanced individual use
$200/mo
Businessteam
Starts at 2 seatsExtra credits availableCodex seats supported
$25/seat/moannual billing

Gemini Code Assist

Google stack

Individuals can start free, then teams move into Standard or Enterprise with a clear trial path.

Models
  • Gemini 2.5
  • Gemini Code Assist
Individualsfree
Personal developersGemini Code AssistTry before buying
$0
Standard
Team managementSeat-based buying30-day trial up to 50 seats
$19/seat/moannual; $22.80 monthly
Enterpriseadvanced
Stronger governanceLarger orgsEnterprise purchase flow
$45/seat/moannual; $54 monthly

Windsurf Plans

Agent workflow

Free, Pro, Teams and Max all revolve around Cascade prompt credits and agent-heavy workflows.

Models
  • Cascade
  • Claude
  • GPT
  • Gemini
Procore
500 prompt credits / monthHeavy personal useCredits do not roll over
$15/mo
Teams
Team seatsCentral adminSeat-based credit allocation
$30/seat/mo
Maxheavy
Highest individual tierLargest allowanceFor very intense workflows
$200/mo

FAQ

The common questions people ask before paying for a coding plan.

Which models are included in GLM Coding Plan?

Zhipu GLM Coding Plan currently supports GLM-5, GLM-4.7 and GLM-4.6. New models are expected to be added over time.

What is a coding plan?

A coding plan is a subscription built for AI coding workflows. You subscribe monthly or yearly and use supported models from tools like Claude Code, Cursor or Cline.

Which coding tools are usually supported?

Most plans support mainstream tools such as Claude Code, Cursor, Cline, Roo Code, Kilo Code and OpenCode. Exact support differs by provider.

What does a 5-hour limit mean?

Many providers use a rolling 5-hour window. Your quota recovers automatically over time instead of resetting at a fixed clock hour.

How much quota does one prompt consume?

One prompt inside an AI coding tool can trigger multiple model calls for search, analysis and generation, so the real cost depends on the provider's metering method.

Which plan has the best value?

MiniMax Starter is the cheapest entry. Volcengine is strongest for model variety. Zhipu stands out if MCP matters. Kimi adds broader member perks.

Can I get a refund?

Refund policy differs by provider, but most plans are not refundable after purchase. The safest path is to start from the cheapest tier and upgrade after validation.