ZH
2026 · Updated regularly

AI Coding Plans
Compared Side by Side

Claude Code · GLM · MiniMax · Kimi · Qwen
Pricing, models, usage limits, and supported tools — find the plan that fits your workflow

6
Platforms
$18
Lowest / mo
10+
AI Models
20+
Dev Tools

Quick Comparison

Entry-level plans at a glance — find your starting point

Scroll horizontally for full table
Platform Starting Price Core Models Usage (5h window) Tiers Highlight
Claude Code
 $20/mo Opus 4.6, Sonnet 4.6, Haiku 4.5 Usage limits apply 3 tiers Gold Standard
GLM (Z.ai)
 $18/mo GLM-5.1, GLM-5-Turbo, GLM-4.7, GLM-4.5-Air ~80 prompts 3 tiers Free MCP Tools
MiniMax
 $10/mo M2.7, M2.7-highspeed 600 req/5h 6 tiers 100+ TPS
Kimi
 ¥49/mo Kimi K2.6 Weekly quota 4 tiers Membership
Qwen (Alibaba)
 ¥200/mo Qwen3.6-Plus, Kimi-K2.5, GLM-5, M2.5 6,000 requests 1 tier Limited stock
Xiaomi MiMo
 ¥39/mo mimo-v2.5-pro · mimo-v2.5 · mimo-v2-flash Credits-based, 0.8x off-peak 4 tiers OpenAI + Anthropic API

Detailed Plans

Deep dive into each platform's pricing and features

Claude Code Gold Standard

Anthropic's flagship AI coding agent — terminal, IDE, web, and desktop
Models: Opus 4.6 · Sonnet 4.6 · Haiku 4.5
ProEntry
Includes Claude Code & Cowork Unlimited projects Memory across conversations
$20/mo
$17/mo billed annually
Max 5xPopular
5× Pro usage Higher output limits Priority access
$100/mo
Max 20xPower User
20× Pro usage Early access to features Priority at peak times
$200/mo
Available On
Terminal CLIVS CodeJetBrainsWebDesktop AppSlack
Opus 4.6 Extended Thinking Web Search Connectors / MCP
View Plans

GLM Coding Plan Free MCP

By Zhipu AI (Z.ai) — GLM-5.1 on par with Claude Opus 4.6 on LMArena Code, Singapore-based service
Models: GLM-5.1 · GLM-5-Turbo · GLM-4.7 · GLM-4.5-Air
LiteBudget
~80 prompts / 5 hours ~400 prompts / week Vision + Web Search + Zread MCP included
$18/mo
$16.2/mo from 2nd quarter
ProBest Value
~400 prompts / 5 hours ~2,000 prompts / week For high-frequency complex projects
$72/mo
$64.8/mo from 2nd quarter
MaxMax Usage
~1,600 prompts / 5 hours ~8,000 prompts / week First access to new models · Priority access
$160/mo
$144/mo from 2nd quarter
Supported Tools
Claude CodeRoo CodeKilo CodeClineOpenCodeCrushGooseOpenClawCursor20+
Vision MCP Web Search MCP Web Reader MCP Zread MCP GLM-5.1 SOTA Referral up to 20% Credits
Subscribe

MiniMax Token Plan 100+ TPS

Mini price, Max performance — ultra-fast inference, 2 months free on annual plans
Models: M2.7 · M2.7-highspeed
Standard Plans (M2.7 ~50 TPS)
StarterEntry
600 req / 5 hours Lightweight workloads
$10/mo
$100/yr (save $20)
Plus
1,500 req / 5 hours 2.5× Starter usage
$20/mo
$200/yr (save $40)
Max
4,500 req / 5 hours 7.5× Starter usage
$50/mo
$500/yr (save $100)
High-Speed Plans (M2.7-highspeed ~100 TPS)
Plus HS
1,500 req / 5 hours ~100 TPS high-speed inference
$40/mo
$400/yr (save $80)
Max HSTop Pick
4,500 req / 5 hours ~100 TPS high-speed inference
$80/mo
$800/yr (save $160)
Ultra HSFastest
30,000 req / 5 hours 50× Starter usage
$150/mo
$1,500/yr (save $300)
Supported Tools
OpenClawClaude CodeRoo CodeKilo CodeClineCodex CLIOpenCodeCursorTRAEGrok CLI
Image Understanding MCP Web Search MCP 100+ TPS Referral 10% Back
Subscribe

Kimi Code Plan Membership Perks

By Moonshot AI — Kimi K2.6 model, 4 tiers, weekly quota refresh, AI website/docs/PPT/spreadsheet included
Model: Kimi K2.6
AndanteDaily Use
Dedicated quota · Weekly refresh Flagship model early access
¥49/mo
Annual ¥39/mo (¥468/yr)
ModeratoRecommended
Larger quota · Weekly refresh Multi-device shared quota Claude Code / Roo Code support
¥99/mo
Annual ¥79/mo (¥948/yr)
AllegrettoPro
20x Kimi Code quota · Kimi Claw Concurrency up to 30
¥199/mo
Annual ¥159/mo (¥1,908/yr)
AllegroUltra
60x Kimi Code quota · Kimi Claw + Agent Swarm Save ¥1,680/yr on annual plan
¥699/mo
Annual ¥559/mo (¥6,708/yr)
Supported Tools
Kimi CLIClaude CodeRoo CodeOpenCodeKimi Code for VS Code
Kimi K2.6 Weekly Refresh Kimi Claw AI Website/Docs/PPT/Sheet Save up to ¥1,680/yr
Get Started

Qwen Coding Plan Limited Stock

By Alibaba Cloud — Qwen series + third-party models. Pro tier only, limited daily stock — restocked at 09:30 BJT
Models: Qwen3.6-Plus · Kimi-K2.5
GLM-5 · MiniMax-M2.5
Qwen3-Coder-Next + 4 more
ProOnly Tier
6,000 requests / 5 hours 45,000 requests / week 90,000 requests / month
¥200/mo
Limited stock · restocked 09:30 BJT daily
⚠️ Note: One prompt in a coding tool typically triggers 5–30 model calls. Quota resets on a rolling 5-hour, weekly, and monthly cycle. No refunds.
Quota Reset Rules
5h rolling refreshWeekly resetMonthly reset on billing date
Qwen3.6 Series Third-Party Models Limited Stock
Subscribe

Xiaomi MiMo Token Plan Claude + OAI API

8 models, one subscription. Off-peak discount (0.8×), compatible with Anthropic & OpenAI API formats
Models: mimo-v2.5-pro · mimo-v2.5
mimo-v2-omni · mimo-v2-flash
TTS series (free for now)
LiteStarter
4.1B Credits / month Off-peak × 0.8 All 8 models
¥39/mo
Annual ¥34.32/mo (¥411.84/yr)
StandardAdvanced
11B Credits / month Off-peak × 0.8 All 8 models
¥99/mo
Annual ¥87.12/mo (¥1,045.44/yr)
ProBest
38B Credits / month Off-peak × 0.8 All 8 models
¥329/mo
Annual ¥289.52/mo (¥3,474.24/yr)
MaxPower
82B Credits / month Off-peak × 0.8 All 8 models
¥659/mo
Annual ¥579.92/mo (¥6,959.04/yr)
💡 Credits-based billing: deducted by actual token usage · Off-peak (0:00–8:00 BJT) consumption × 0.8 · First purchase 12% off · New users: 23% off on first auto-renewal · No refunds
Supported Tools
Claude CodeOpenClawOpenCodeKiloCodeRoo CodeClineCodexCherry StudioZedCodeBuddy+more
mimo-v2.5-pro Credits billing Off-peak 0.8× OpenAI + Anthropic API Free TTS
Subscribe Now

Frequently Asked Questions

Common questions about AI Coding Plans

What is a Coding Plan?
A Coding Plan is a monthly subscription offered by various AI platforms that gives you access to their models inside coding tools like Claude Code, Cursor, Cline, and more. Instead of paying per token, you pay a flat fee and get a generous usage quota that refreshes on a rolling basis.
Which coding tools are supported?
Most plans support Claude Code, Cursor, Cline, Roo Code, Kilo Code, and OpenCode. GLM and MiniMax support 10–20+ tools. Kimi works primarily through Kimi CLI, Claude Code, and Roo Code. Claude Code's own plan works across terminal, VS Code, JetBrains, web, and desktop.
What does "5-hour rolling limit" mean?
Most platforms use a rolling 5-hour window to manage usage. For example, MiniMax Starter allows 600 requests per 5-hour window. As time passes, your oldest usage "falls off" and quota frees up automatically — you don't have to wait for a fixed daily reset.
How many model calls does one prompt use?
In coding tools, a single user prompt typically triggers 5–30 model calls behind the scenes, as the tool performs code analysis, search, generation, and validation steps. Different platforms count usage differently — some count "prompts" (user messages), others count raw API requests. Check each platform's docs for specifics.
Which plan offers the best value?
It depends on your needs. For the lowest entry price, GLM Lite at $18/month is a strong starting point with free MCP tools included. For a balance of price and speed, MiniMax Starter at $10/month gives you 600 requests per 5 hours. For the best model quality, Claude Code Pro at $20/month gives you access to Opus 4.6. For multi-model flexibility, Volcengine Ark lets you switch between 10+ models.
Can I get a refund?
Refund policies vary by platform. Most subscriptions are non-refundable once activated. We recommend starting with a lower tier to test the service before committing to a higher plan.