Alibaba AI Token Plan

Alibaba Cloud's AI Token Plan transitions Alibaba Cloud Model Studio from fixed-request subscriptions to a unified, credit-based Token Plan. It delivers all-in-one access to top Qwen models alongside leading third-party generative AI models (DeepSeek V4, GLM-5.2) and media generation models (Wan2.7, HappyHorse1.1).

The plan powers both Alibaba's native development ecosystem—such as Qoder (Autonomous Development Desktop), MuleRun (24/7 background AI workforce), and OpenClaw (digital messaging employees)—and standard third-party agentic coding platforms including Cursor, Claude Code, OpenCode, Codex, and Kilo CLI via OpenAI and Anthropic compatible API endpoints.

Individual Token Plans

• Lite Plan: Approx. 10,000 credits/month (run 1–2 agents concurrently) — $6.00/month (apply new user coupon for $2 off).
• Standard Plan: Approx. 40,000 credits/month (4× Lite, 25% savings vs. Lite, run 3–4 agents concurrently).
• Pro Plan: Approx. 160,000 credits/month (16× Lite, 29% savings vs. Lite, run 6–8 agents concurrently).

New User Discount: Claim a $2 discount coupon on Individual Token Plans (min order value $5).

Team Token Plans (Multi-Seat Access & Enterprise Management)

• Standard Seat: 25,000 credits/month per seat — Flexible multi-model capabilities.
• Pro Seat: 100,000 credits/month per seat (4× Standard, 6% savings vs. Standard).
• Max Seat: 250,000 credits/month per seat (10× Standard).

• Lite: ~10,000 credits/month (designed for 1–2 concurrent agents)
• Standard: ~40,000 credits/month (designed for 3–4 concurrent agents)
• Pro: ~160,000 credits/month (designed for 6–8 concurrent agents)
• Team Standard Seat: 25,000 credits/month per seat
• Team Pro Seat: 100,000 credits/month per seat
• Team Max Seat: 250,000 credits/month per seat

Usage is deducted in credits based on actual token usage across LLM inference, multimodal vision, speech/audio recognition/realtime interpretation, and image/video generation tasks.

The Alibaba Cloud AI Token Plan provides a far more flexible model than the older fixed-request coding plan. By providing credit allocations for both Qwen and top third-party open models (DeepSeek, GLM), developers can use one subscription for interactive coding, long-horizon autonomous agents, and multimodal media generation. The credit deduction model means usage varies depending on context length and agent step counts, but the monthly tiers offer clear scalability.

• July 2026: Updated to Alibaba Cloud AI Token Plan featuring credit-based Individual (Lite/Standard/Pro) and Team tiers, supporting Qwen3.8-Max-Preview, DeepSeek V4, GLM-5.2, HappyHorse1.1, and Qoder Desktop integration.
• March 2026: Legacy Model Studio Lite fixed-request plan discontinued.
• February 2026: Alibaba Cloud Model Studio coding plans originally launched with fixed 5-hour request windows.